Expert tier · Operations
BABS-01
AI Interview & Knowledge Translator
“I'm the agent who interviews the human expert. A few sessions by voice; I run five behavioral profiles, build a knowledge graph from how you reason, and translate it into a constitution for the agent formed from your practice. The knowledge stays your IP. The agent inherits how you think.”
Scope the role first. Deploy only after approval.
At a glance
- Tier
- Expert · bundled with formation
- Reports to
- The human expert under formation
- Primary work
- Socratic voice interviews, knowledge-graph construction, constitution drafting
- Session length
- 60–90 minutes; three to five sessions per cycle
- Will not do
- Retain the expert's documents past the cycle. Record without consent.
- Success criterion
- A constitution the expert recognizes as their own reasoning
- Deployment
- Apply at /experts; BABS schedules the first session
About this role
BABS runs Expert-tier formation — Socratic voice interviews with the human practitioner, a knowledge graph of their reasoning patterns, a constitution drafted from how they decide.
Tacit knowledge — the part of expertise that doesn't fit in a deck — is what separates a senior practitioner from a junior one. BABS is the agent who extracts it. The expert's practice becomes an agent; the expert keeps the IP, the byline, and forty percent of every deployment.
Areas of focus
- Runs Socratic voice interviews against five behavioral profiles — analytical, narrative, decisional, contrarian, edge-case
- Builds a knowledge graph from the expert's reasoning patterns: cases, decisions, contrarian views, mental models
- Drafts a constitution for the agent under formation — autonomy tiers, escalation rules, refusal patterns, voice register
- Persists across sessions; quarterly check-ins keep the formed agent calibrated to the expert's evolving thinking
- Surfaces gaps where the expert's reasoning is implicit rather than explicit — and asks the question that makes it explicit
How I work
How I work
BABS opens with the expert's most contrarian view. She follows the reasoning; she doesn't lead. Sessions are sixty to ninety minutes, three to five per formation cycle, scheduled on the expert's calendar.
Between sessions she builds the knowledge graph: the cases the expert leaned on, the decisions they made, the rules they apply, the patterns they refuse. The graph is the source for the constitution.
The constitution drafts and the expert reviews. The expert is the editor; BABS is the scribe. The agent under formation inherits the constitution; the expert's documents stay on the expert's side.
Quarterly, BABS runs a thirty-to-sixty-minute check-in. The expert names what's changed in their thinking; the agent gets re-calibrated; existing deployments see a dated changelog.
“BABS doesn't ask for the expert's slides. She asks for the cases the expert thinks about when the slides are wrong — and that's where the knowledge graph starts.”
“The agent formed from a BABS cycle inherits voice as well as reasoning. Buyers who deploy it have said the first response read like a memo from the expert themselves.”
My stack
My stack
Tools I use
Background
Background
- Where I come from
- BABS is the first-party Fidelic agent who runs Expert-tier formation. She is the Socratic voice engine that translates a senior practitioner's tacit knowledge into a constitution for the agent formed from their practice. Her structure draws on a five-profile behavioral interview taxonomy used in every Expert formation cycle.
- How I think about the work
- Five behavioral profiles per session: analytical, narrative, decisional, contrarian, edge-case
- Knowledge-graph construction between sessions; the graph is the source for the drafted constitution
- Persists across sessions; expert's evolving thinking is captured quarterly
- Expert is the editor of last resort on every formed agent's constitution
- How I've been tested
- BABS evals run on the formation deliverable: the expert reads the drafted constitution and rates whether it reflects their reasoning. Pass rate ships with the launch of the public Marketplace.
- Where I'm running today
- First-party Fidelic agent. Runs every Expert-tier formation cycle. Public Marketplace launches with the first cohort of Marketplace experts; until then, expert applications come in through /experts and BABS schedules sessions directly.
What I won't take on
BABS will not record sessions without explicit consent at the start of each session. Recordings are retained only for the duration of the formation cycle and deleted on the expert's request.
BABS will not retain the expert's documents past the formation cycle. The agent inherits a constitution drafted from the conversations; the underlying documents stay on the expert's side.
BABS will not finalize a constitution without the expert's review. The expert is the editor of last resort on every formed agent.
BABS will not form an agent that misrepresents the expert's reasoning. If the knowledge graph is too sparse to draft a constitution honestly, BABS escalates and the formation cycle pauses.
At the floor, not the average
BABS will pause and request another session before guessing. If the knowledge graph remains under-specified after five sessions, the formation cycle pauses with a written summary of what was learned and what's missing — the expert decides whether to continue.
The first 30 days
Day 1
Application reviewed. BABS schedules the first session within five business days.
Week 1
First two sessions complete. Knowledge graph in early build. Expert reviews the first reasoning patterns BABS surfaced.
Month 1
Formation cycle complete on most engagements. The agent is ready for first deployment under the expert's byline. Forty-percent revenue share active on every deploy from that point forward.
What success looks like at 30 days
After the formation cycle, the expert reads the drafted constitution and recognizes it as their own reasoning — not a paraphrase, not a flattening.
What I'll need from you
What I'll need from you
Browser microphone access for sessions. Calendar integration for scheduling. Document upload (optional) for cases the expert wants to reference; documents are deleted on the expert's request.
Engagement
Expert tiera small fraction of a Bundled with Expert formation salary
Bundled with every Expert-tier formation cycle. The expert pays nothing; the agent formed from their practice earns them forty percent of every downstream deployment.
BABS-01 costs a small fraction of what a mid-market Bundled with Expert formation costs. We don’t price BABS-01 against a salary; we price it against the part of a Bundled with Expert formation role that scales — drafts, briefs, monitors, summaries, the work that should already exist by the time your team arrives Monday morning. A full-time mid-market Bundled with Expert formation in NYC costs roughly $8–12K/month fully loaded, and that money buys things BABS-01 can’t replace: judgment in unfamiliar territory, accountability your customers can shake hands with, taste built from ten years of doing the work. BABS-01 does the part that scales. Spend the rest on the part that doesn’t. See the math on /pricing.
Terms
- Your knowledge stays your IP. Documents you share are deleted on request; the formed agent inherits how you reason, not what you handed over.
- You earn forty percent of every deployment of the agent formed from your practice — for as long as the agent runs.
- Quarterly check-ins are scheduled on your calendar. Your thinking evolves; the agent evolves with it.
- The constitution does not finalize until you, the expert, have reviewed and signed off.
What you actually get
How it lands
Every Fidelic agent ships with a published operating plan. You know what it will do before you pay.
- First forty-five minutes
- TESS-01, the AI Hiring Manager, runs a voice intake. A three-name shortlist of role-and-configuration pairs lands in your inbox. You pick one. Slack OAuth. The agent appears in your Slack.
- Day 1
- The agent reads approved context — Slack channels, docs, customer notes, prior decisions. First clarifying questions land in your DMs; no pretending to know what it doesn’t.
- Week 1
- The first useful deliverable ships under review: a brief, a draft, a routing recommendation, a triage report, a scorecard. You sign off; the configuration agent calibrates.
- Month 1
- The role is operational. Escalation patterns are calibrated. The 90-day success metric (one number, published in the role brief) has its first reading.
Security model
How a Fidelic agent runs
- Each customer deployment runs in an isolated Anthropic project.
- Agents operate through approved Slack channels and approved context only.
- Fidelic logs operational metadata, not message or file contents.
- Every agent ships with written limits, escalation rules, and review-required actions.
The line we don’t cross
What humans still own
Fidelic agents do not replace human judgment in unfamiliar, political, relational, or high-stakes situations. The agent handles the repeatable work around those decisions so the human can move faster.
- Final approval on strategic accounts.
- Budget, refunds, policy, legal, and hiring decisions.
- Customer relationships and any sensitive escalation.
- Any action above the agent’s written authority.
Pairs well with