LIVE AUDITSee how your business can save money and time.
AUTOMATIONS · KNOWLEDGE · AI

Internal knowledge base AI automation.

Employee asks once; AI retrieves from docs, people, and live systems with permission-aware filtering. Every answer cited to source. Read-only tool calls against Salesforce, GitHub, JIRA. Acknowledges gaps instead of hallucinating. KB-gap loop captures human-answered questions and turns them into permanent knowledge. The 'who-do-I-ask' tax drops from 4–8 hours to under 30 minutes per knowledge worker per week.

TYPICAL SAVINGS $120K–$840K/yr
DEPLOY TIME 6–10 weeks
COMPLEXITY Tier 3
MONTHLY COST $420–$2,200/mo
WHAT THIS IS

A real internal AI has four jobs.

Most internal AI deployments are ChatGPT with company logo and a doc upload. The result is confidently wrong answers about company policy, leaked confidential data because permissions weren't enforced, and an assistant that nobody trusts within a quarter. The job of a real internal AI is to be permission-aware first, source-grounded second, and humble about gaps third — answers cited to specific sources the requester can verify, gaps acknowledged instead of fabricated, sensitive data never crossed.

Four jobs. One: classify the question by intent and likely source — docs-resolvable (policy, runbook), people-resolvable (who-owns-what), systems-resolvable (live data from Salesforce/GitHub/JIRA), or multi-source. Two: filter sources by employee permissions before retrieval, never after generation. Engineering employee asking about salary band data sees engineering policies but not Comp's confidential band tables. Three: retrieve from the right source layer — RAG over indexed docs with citations + staleness signals; people-graph for ownership lookup with intro DM drafting; tool-calls for read-only system queries with audit trails. Four: synthesize multi-source answers with conflicts flagged; honestly acknowledge gaps when no confident answer exists; capture human-resolved questions back into the KB so the next person gets a confident answer.

Done right, your knowledge worker loses less time to 'who do I ask' wandering, your senior people stop being asked the same 12 questions weekly, and your KB compounds quality as gaps get closed by the loop. Done wrong, you ship a confident-without-competence box that confidently misquotes policy, leaks salary data through ACL bypasses, and either gets shut down by Legal or quietly ignored by employees within two quarters.

BEFORE

Slack search + DM the senior person

Engineer needs to know the security review SLA. Searches Slack — finds 14 conversations with conflicting answers from 2019, 2022, 2024. Searches Notion — finds 3 docs that disagree. Gives up, DMs the security lead. Security lead gets the same question 6 times this week. Engineer waits 3 hours for response. Security lead's deep work fragmented across 12 'quick questions' per day. Total cost: 18 minutes of engineer time + 12 minutes of security lead time per question, multiplied across the org.

AFTER

Permission-aware AI with cited answers

Same engineer asks the assistant. AI classifies as policy-resolvable. Permission filter confirms engineer can see security policies (yes). RAG retrieves the current SLA doc (last updated 6 weeks ago, owned by security lead). AI returns: 'Security review SLA is 5 business days for standard requests, 2 business days for P0 launches. Source: Security Review Process v3.2 (link). Last updated by Maria 6 weeks ago.' Engineer reads in 90 seconds, has the answer with confidence. Security lead's question volume drops 60% in 90 days.

FIT CHECK

Who this is for, who it isn't.

Internal knowledge AI pays back fastest for businesses with 200+ employees, mature documentation in modern doc tools (Notion/Confluence/Drive), and visible 'who-do-I-ask' friction. Below 100 employees, search is fine. Below 50 employees, DMs work. The threshold is when senior people are being interrupted dozens of times daily by routine questions.

HIGH LEVERAGE FOR

Build this if any of these are true.

  • You have 200+ employees and your senior people complain about being asked the same questions repeatedly. That's the time being recovered.
  • Your existing doc search is broken — Slack search returns conflicting old conversations, Notion search misses cross-workspace content, Drive search depends on exact filenames.
  • Your knowledge sits across 3+ doc systems (Notion + Drive + Confluence + Slack canvases) and employees can't find anything because nothing knows where else to look.
  • You have an SSO identity provider (Okta, Azure AD, Google Workspace) with reliable group membership. Without strong identity, permission-aware retrieval can't be built safely.
  • You have a People-team or IT partner who can own permission-model design and ongoing audit. This isn't a one-team build.
SKIP IF

Skip or wait if any of these are true.

  • You're under 100 employees. The marginal value doesn't justify the build complexity at low headcount.
  • Your documentation is genuinely scattered and outdated. Fix the docs first; AI on top of bad docs amplifies the bad. The KB-gap closure loop only helps marginally if 60% of your existing docs are stale.
  • Your permission model is loose (everyone can see everything). Either you accept the data leakage risk or you fix the permission model first.
  • You're regulated industry where AI access to internal data has specific compliance requirements (some financial services, healthcare with HIPAA constraints, government contractors with FISMA). Build the compliance frame first.
  • You're hoping employees ask the AI in lieu of asking experts. They won't, and that's fine. The good version handles routine questions; expert questions still go to experts.
Decision rule: If you have 200+ employees, mature docs in modern tools, working SSO, and a People-team partner for permission ownership, this is one of the highest-leverage Tier-3 knowledge automations. Skip if your docs need cleanup first or your permission model needs tightening.
THE HONEST MATH

What this saves, by the numbers.

The savings come from three sources, in order. Knowledge worker time recovered from 'finding the answer' wandering (the largest line at scale — every knowledge worker loses 4–8 hours per week to it). Senior expert capacity preserved as routine questions get answered without their interruption. New-employee ramp time compressed from clarity gains. Most teams see 1.5–2× the conservative numbers below by year two.

UNIVERSAL FORMULA
(Knowledge workers × hrs saved per week × loaded hourly cost × weeks/yr) + (senior expert hours preserved × hourly cost) + (new-hire ramp compression × salary)
Hours saved per worker per week = roughly 2–4 hours from faster knowledge access. Senior expert preservation = avoided interruption time on the senior people who get asked the same things repeatedly. Ramp compression = days of full productivity recovered for new hires (typical: 5–10 days off the typical 90-day ramp).
SMALL OPERATOR
250 employees · 30 senior experts · $100K avg loaded cost
$120K
per year saved
WORKER TIME: 250 × 2hr/wk × 48 × $50 = $1.2M (gross) EXPERT PRESERVATION: 30 × 4hr/wk × 48 × $90 = $518K RAMP: 50 hires × 5 days × $385 = $96K MINUS BUILD + TOOLING: $84K NET YEAR 1: ~$120K MATURE YEAR 2+: ~$320K
MID-SIZE
1,200 employees · 150 experts · $130K avg
$420K
per year saved
WORKER TIME: 1,200 × 3hr/wk × 48 × $65 = $11M (gross) EXPERT PRESERVATION: 150 × 5hr/wk × 48 × $120 = $4.3M (gross) RAMP: 240 hires × 8 days × $500 = $960K MINUS TOOLING + OPS: $144K NET YEAR 2+: ~$420K conservative
LARGER SCALE
5,000 employees · 600 experts · $160K avg
$840K
per year saved
WORKER TIME: 5,000 × 4hr/wk × 48 × $80 = $77M (gross) EXPERT PRESERVATION: 600 × 6hr/wk × 48 × $140 = $24M (gross) RAMP: 1,000 hires × 10 days × $615 = $6.15M MINUS TOOLING + OPS: $300K NET YEAR 2+: ~$840K conservative
What's not in those numbers: Compound knowledge effects (the longer the system runs, the more KB gaps close and the higher the helpful-answer rate climbs), reduced senior-expert burnout from interruption load, faster cross-team collaboration as expertise becomes findable, and second-order benefits to org learning as historical decisions remain accessible to future employees. Most teams see 1.5–2× the conservative numbers above by year two.
HOW IT WORKS

The architecture, end to end.

Internal AI architecture has a single trunk (question intake, AI classify, permission filter) feeding 3 source lanes. Docs lane runs RAG over indexed sources with citation-grounded answers. People lane runs expertise + ownership lookup with intro DM drafting. Systems lane runs read-only tool calls against live data with full audit trails. All three lanes converge at a synthesize step that combines multi-source answers with conflicts flagged. Confident answers deliver with feedback capture; gaps get acknowledged honestly and routed to a KB-closure loop that captures human-resolved answers back into the KB. Click any node for the architectural detail; click a path label to highlight one route.

+ Click any node to expand. Click a path label below to highlight one route through the graph.

DOCS PEOPLE SYSTEMS ANSWERED KB GAP CLOSE
TRUNK · CLASSIFY + PERMISSION
TRIGGER
Employee question received

Slack/Teams/widget. SSO-authenticated identity, not user-claimed.

AI
AI / CLASSIFY
Identify intent + source signals

Docs / people / systems / multi-source. Each tagged with confidence and sensitivity.

03
PERMISSION
Filter sources by employee access

Filter before retrieval, not after generation. Bypassing this is how internal AI ships data leakage.

PATH · DOCS
📄
DOCS
RAG over indexed sources

Notion/Confluence/Drive/SharePoint/wikis. Staleness signals attached to each chunk.

📄↓
DOCS
Cited answer + source links

Every claim cited. Stale-doc warnings inline. Conflicting sources flagged, not silently merged.

PATH · PEOPLE
👥
PEOPLE
Expertise + ownership lookup

HRIS + GitHub + doc auth + assignments. Reduces "who do I ask" tax (4–8 hr/wk per worker).

👥↓
PEOPLE
Intro + context handoff

AI drafts intro DM with context attached. Saves 8–12 min context-loading per routed question.

PATH · SYSTEMS
SYSTEMS
Tool-call query against live data

Read-only only. Write actions go through dedicated tool UI with explicit intent.

⚙↓
SYSTEMS
Format response + audit trail

Sensitive queries get extra audit + manager visibility. Audit trail enables compliance.

SYNTHESIZE
SYNTHESIZE
Combine sources into single answer

Multi-source answers feel different from a search box. Conflicts surface, never silent merge.

OUTCOME · ANSWERED
ANSWERED
Deliver + capture feedback

Helpful yes/no/partially. Quality data tunes the model. Target: 85%+ helpful-rate at 90 days.

✓✓
SUCCESS
Cache + reuse for similar Qs

Semantic cache cuts AI cost 30–50% as common questions repeat. Quality + cost improve.

OUTCOME · KB GAP
?
KB GAP
Acknowledge gap + escalate

Better to admit uncertainty than confidently hallucinate. Hardest discipline in internal AI.

?↓
KB GAP
Capture answer + close the gap

AI drafts KB article from human's answer. Next person gets confident answer. System learns.

TOOLS YOU'LL USE

Stack combinations that actually work.

Three stack combinations cover most builds. The decision usually comes down to your existing doc tools and how custom you need the AI layer. Glean and Hebbia dominate enterprise turnkey. Notion AI + Slack covers mid-market natively. Custom builds with Claude or GPT offer the most flexibility but require engineering capacity.

COMBO 1
Glean + Slack + Okta
$1,200–$2,200/mo

Tradeoff: The enterprise turnkey stack. Glean handles indexing across 100+ doc sources with native permission inheritance; Slack/Teams as the surface; Okta provides identity. About $1,500/mo all-in for 500-employee org. Best for $50M+ revenue with mature doc ecosystem. Hits a ceiling on Glean's per-seat pricing past 2,000 employees.

COMBO 2
Notion AI + Claude + Slack
$540–$1,200/mo

Tradeoff: The Notion-native stack. Notion AI handles in-Notion search; Claude with Pinecone extends RAG to Drive, GitHub, Slack canvases; Make orchestrates cross-system flows. Best for Notion-primary companies in the $5M–$30M revenue range. Lower per-seat cost than Glean; less mature multi-system handling.

COMBO 3
Custom: Postgres + Pinecone + Claude + Slack
$420–$840/mo

Tradeoff: Most flexible. Postgres holds metadata; Pinecone the vector store; Claude composes; custom Slack app + web widget for interfaces. Best for technical companies with engineering capacity and unusual integration needs (proprietary tools, custom permission models). Highest build complexity; most flexible permission-model design. Worth it past 1,000 employees.

MINIMUM VIABLE STACK
Notion AI + Slack channel

Cheapest viable. Notion AI ($10/seat/month) for in-Notion search + Slack channel where employees can ask questions and Notion AI bot responds. Skip the multi-source layer initially. Validates whether employees will actually use an AI assistant before investing in the full pipeline. About $0 above existing Notion. Builds in 1 week.

PRODUCTION-GRADE STACK
Glean + Okta + Claude + Slack + audit dashboard

Production stack for $50M+ revenue with 500+ employees. Glean ($30–$50/seat at scale), Okta with deep group inheritance, Claude Opus for extended composition ($150–$400/mo), Slack with channel + DM surfaces, custom audit dashboard for security + compliance review. About $1,800–$3,200/mo all-in. Adds the permission-audit infrastructure, KB-gap closure loop, and quarterly tuning rhythm.

THE BUILD PATH

How to actually build this.

Six steps from zero to a production internal AI. The biggest mistake teams make is shipping cross-source AI before the permission model is bulletproof — bypassing permission-aware retrieval ships data leakage that surfaces at exactly the wrong moment.

01

Lock the permission model

Document every doc-source's permission model and how it inherits to retrieval. Notion permissions = workspace + page-level. Drive permissions = file + folder + group inheritance. Confluence permissions = space + page. Slack permissions = channel membership + DM history. Map each to your SSO group structure. Validate: can engineer A see HR's confidential salary docs? (Should be no.) Validate before building retrieval, not after.

What's at risk: Permission model assumed to work. Hidden document shares (random people getting access via 'anyone with the link' shares from years ago), legacy folder permissions, deleted-but-cached docs all leak data through naive retrieval. Audit explicitly; trust nothing.
ESTIMATE 7–11 days
02

Build the indexed corpus + retrieval

Index every doc source into the vector store with permission metadata attached to each chunk. Re-index on doc updates (most platforms support webhook-triggered reindex). Test retrieval against 100 sample queries: do you get the right docs back? Do permission filters work as expected? Validate that requesting employee A doesn't get back chunks they couldn't see in the source system.

What's at risk: Stale index. Doc updated yesterday but index reflects the version from last month. Build webhook-driven reindex with a fallback nightly full reindex. Critical for policy docs where outdated answers cause real problems.
ESTIMATE 8–12 days
03

Build AI classification + permission gate

Wire AI classification: docs / people / systems / multi-source intent identification. Critical: permission filter applies to source selection BEFORE retrieval, not to chunks AFTER. Employee asks 'what's John's salary?' — permission filter says no comp-data access for this employee, retrieval is never executed. The filter is upstream, not a post-processing step.

What's at risk: Permission applied as post-filter on retrieval results. Means retrieval already touched data the employee shouldn't see, and the model could leak fragments. Hard architectural rule: permission gates retrieval inputs, not retrieval outputs.
ESTIMATE 5–8 days
04

Build the three source lanes

Docs lane: RAG with citation-grounded answers, staleness signals, conflict flagging. People lane: expertise lookup from HRIS + GitHub + doc-authorship graph; intro DM drafting. Systems lane: read-only tool calls against Salesforce, GitHub, JIRA, expense tools, with deep-links and audit trails. Build them in usage-frequency order — docs first (highest volume), systems second, people third.

What's at risk: Tool calls without permission scoping. AI invokes 'GET /salesforce/account?id=X' for an employee who doesn't have access to that account. Each tool integration must verify the employee's permission against the underlying system before executing the query.
ESTIMATE 11–17 days
05

Build synthesis + KB-gap loop

Synthesis layer: combine multi-source answers with conflicts flagged. Confidence below threshold triggers gap-acknowledgment ('I don't have a confident answer; here's what I found') instead of fabrication. Gap queue: questions the AI couldn't confidently answer get logged. When the human-asked-person answers, AI offers to draft a KB article. SME reviews and publishes. The gap loop is what makes the system learn.

What's at risk: Gap closure loop ignored by SMEs. If SMEs don't review and publish the AI-drafted KB articles, the gap stays open forever. Build review SLA into operating rhythm; track gap-closure rate as a quarterly metric.
ESTIMATE 5–8 days
06

Audit + observability

Every query logged: who asked what, what sources were used, what was returned, was it confident, did the user mark it helpful. Sensitive-data queries (compensation, customer PII, M&A topics) get extra audit treatment. Quarterly access audit: are permission filters working as expected? Random sample of queries reviewed for false-positive permission grants. Build observability dashboard: helpful rate, gap rate, source distribution, latency, cost.

What's at risk: Sensitive-query audit skipped. Without it, permission bypass goes undetected until exposed. Quarterly review of sensitive-flag queries is non-negotiable. Compliance owner must sign off.
ESTIMATE 4–7 days
TOTAL BUILD TIME 6–10 weeks · 1 builder + 1 IT/identity lead + 1 KB owner
COMMON ISSUES & FIXES

Where this fails in real deployments.

Five failure modes that wreck internal AI in production. Every team that's built this hits at least three of them.

01

Permission bypass leaks salary data

Engineer asks 'how do salary bands work?' AI's RAG retrieves the comp policy doc (which the engineer can see). It also retrieves a compensation-bands spreadsheet from a Drive folder that was shared 'anyone in company' three years ago by an HR person who left. Engineer sees the actual band table for executives. Information that was never supposed to be company-wide is now in this employee's hands.

How to avoid: Permission-aware retrieval audits both current ACLs and historical permission grants. 'Anyone in company' shares from doc systems get reviewed quarterly — does this still belong to be visible to everyone? IT-led audit identifies and fixes legacy over-shares before retrieval indexes them. Sensitive-keyword queries (salary, comp, pay, bonus) get extra-strict filtering and log to compliance review.
02

AI confidently quotes outdated policy

PTO policy changed 6 months ago — employees now get unlimited PTO. Old policy doc (pre-change) still exists in Notion, never archived. AI retrieves both. Without staleness signals, AI returns the old policy confidently. Employee plans vacation around old-policy 15-day cap. Manager corrects them; employee feels the company is disorganized; trust in the AI drops.

How to avoid: Every doc has explicit last-updated metadata in the index. Stale docs (>12 months without verification) flag inline in answers. Conflicting sources within the same answer get flagged: 'two sources disagree on this; here are both.' Quarterly KB hygiene cycle archives outdated docs explicitly. AI never confidently merges conflicting sources into a single answer; it surfaces the conflict.
03

AI hallucinates a policy that does not exist

Employee asks about parental leave. Your company hasn't formalized parental leave policy yet. AI's training implies most companies have one; AI makes up reasonable-sounding details ('most US companies offer 12 weeks'). Employee plans accordingly. Three months later, when actual leave time comes up, the inconsistency surfaces; HR scrambles to define policy under pressure.

How to avoid: Hard constraint: AI never generates content not grounded in retrieved sources. If retrieval returns nothing, AI says 'I don't have a documented policy on this; I'd suggest asking HR' rather than synthesizing one. Confidence below threshold = explicit gap acknowledgment + suggested human contact. The hardest discipline in internal AI is teaching the model 'I don't know' is a valid answer.
04

KB-gap loop generates AI-drafted articles nobody reviews

AI accumulates 200 KB-gap-closure drafts in 6 months. SMEs are too busy to review. The drafts pile up unpublished. The KB doesn't grow. The same questions keep coming back to the AI as gaps. Loop completes the technical cycle but doesn't produce learning.

How to avoid: Review SLA built into operating rhythm: gap-closure drafts get a 14-day SME review SLA. Aging drafts beyond that escalate to manager. Quarterly KB-gap review cadence with documented metrics: gap rate trending, closure rate, KB growth. Without rhythm, the loop fails by neglect, not by design.
05

Cost spikes when usage scales

Pilot launched with 50 power users; AI cost was $400/month. Rolled out to 1,500 employees; cost is now $24,000/month. Procurement and finance scramble to negotiate AI vendor pricing; meanwhile employees are getting throttled or quality degrades to control cost. The pilot's economics didn't predict the full deployment's economics.

How to avoid: Build semantic caching from day one — common questions get cached answers reused across employees. Token-budget per employee (soft limits — 50 questions/day per person, 200 burst). Use cheaper models (Sonnet, GPT-4o-mini) for classification and routing; reserve Opus for the final synthesis only. Quarterly cost review with the caching + tier-routing levers tunable as usage grows.
DIY VS HIRE

Build it yourself, or get help.

This is a Tier-3 build because the permission-model design and audit infrastructure are the hard work, not the AI. Done well, it pays back in months and dramatically improves knowledge worker capacity. Done sloppily, it ships data leakage and confidently-wrong answers at scale.

DO IT YOURSELF

Build it yourself

If you have engineering, IT identity ownership, and KB hygiene partnership.

SKILL Backend engineer + IT/identity lead + KB owner. Comfortable with vector stores, RAG patterns, OAuth/SAML permission inheritance, audit log design. KB owner who can lead the gap-closure cycle and SME review.
TIME 260–400 hours of build over 6–10 calendar weeks, plus 10–14 hours per week of permission auditing, KB hygiene, and AI calibration for the first 90 days.
CASH COST $0 in services. Tooling adds $420–$2,200/mo depending on stack choice and usage volume.
RISK Underestimating permission complexity. Notion's, Drive's, Confluence's, and Slack's permission models all differ. Mapping them to a unified permission-aware retrieval system takes real time. Don't try to bootstrap; get IT identity lead heavily involved from day one.
HIRE A PARTNER

Hire a partner

If knowledge friction is bottlenecking productivity and you can't wait 10 weeks.

SCOPE Full design + build of the internal AI including permission-model audit + design, multi-source indexing with ACL inheritance, AI classification + permission gate, three source lanes (docs/people/systems), synthesis + KB-gap loop, audit infrastructure, observability dashboard, and a 90-day calibration playbook.
TIMELINE 8–12 weeks from contract signed to fully shipped. 30-day stabilization where the partner monitors permission audit findings and tunes thresholds.
CASH COST $48K–$160K project cost depending on doc-source count, permission complexity, and integration depth. Higher end for Glean-led builds with custom systems integrations and multi-jurisdiction compliance.
PAYBACK 5–10 months for most companies with 500+ employees and visible knowledge-friction pain. Faster if senior expert capacity is currently severely constrained.
BEFORE YOU REACH OUT

Want to get in touch with a partner to build this for you? Run the free audit first. It gives any partner the context they need on your business — your stack, your volume, your highest-leverage automation — so the first conversation is about scope, not discovery.

Run the free audit
Decision rule: If you have engineering capacity, an IT identity lead, and existing modern doc tools, build it yourself — the permission audit is your team's to own anyway. If your permission model needs major work or you're under-resourced on KB hygiene, hire a partner. The permission-aware retrieval design is what separates working internal AI from a data-leak liability.
YOUR STACK, AUDITED

Want to know if this is the highest-leverage automation for your business?

Run a free audit. We'll tell you what would save you the most money — even if it isn't this one.

No credit card. No follow-up call unless you ask.