vega
vega securely retains captured ai sessions to build your private work memory and generate performance insights. cross-customer learning is aggregated, anonymized, permissioned, and never resold. how →

Inan Kocatepe

@inan · tryvega.tech/profile/inan

product / vision · Vega Pilot · founding team


8 captured sessions via your connected ai tools

8 approved work signals

75% verification rate

79 avg depth · 73 avg novelty

ai workstyle

deep, novel work — verifies before accepting. shows up across privacy-risk intuition at the product design stage, layered consent architecture thinking, monetization-feature alignment.

archetype
100%
S
A
SA·DV
rank
Apprentice
8 sessions · patterns becoming visible. archetype crystallising.
@inan's ai workstyle

Strategist–Analyst · Deep Verifier

most of your AI sessions sit around decisions, not execution. you weigh tradeoffs, you commit, you move. with a the analyst streak — depth, verification, decomposition.

primary
100%
the strategist
secondary
91%
the analyst
approved signals
8
of 8 captured
archetype radar · all 8 dimensions
SABMEOYD
breakdown
the strategist
100
the analyst
91
the builder
75
the mentor
54
the explorer
53
the operator
53
the synthesizer
22
the drafter
0
interest constellation

where inan's attention goes

each node is a kind of work. size = how much of the captured work it represents. lines connect kinds of work that flow into each other in adjacent sessions.

decide63%debug25%analyze13%
achievements

8 of 12 unlocked

earned by doing real work
First Light
common

your very first captured session

X
Settling In

ten captured sessions

L
Working Pattern

fifty captured sessions

Cross-Checker
rare

verified the AI in 50%+ of sessions

Skeptic
elite

verified in 70%+ of sessions — top decile

Deep Thinker
rare

average depth ≥ 70 across all sessions

Decision-Grade
elite

average decision value ≥ 70

Fresh-Territory
rare

average novelty ≥ 65

Pushback Artist
elite

corrected or challenged the AI in 30%+ of sessions

Polyglot
rare

used three or more distinct AI tools

Range

captured work across five task types

S
On the Record

twenty approved signals on your public profile

connected sources · where inan's captures come from
claude code
5
sessions · last 12h ago
claude.ai
2
sessions · last 2d ago
chatgpt
1
session · last 13h ago
how Inan uses AI
captured sessions
8
top task
decide
63% of sessions
avg depth
79
of 100
avg novelty
73
of 100
avg decision value
79
of 100
verification rate
75%
sessions with a verify step
correction rate
50%
pushed back on AI
approved signals
8
published to profile
workstyle labels · evidence-backed
Strategic decision-maker
63% decide-tasks · avg decision_value 79/100
Deep debugger
25% debug sessions at avg depth 79
Ships what they capture
100% of sessions tied to a shipped outcome
Strong verifier
verified the AI in 75% of sessions
task mix · across 8 sessions
decide
5 · 63%
debug
2 · 25%
analyze
1 · 13%

based on 8 captured sessions

discipline meters
verification rate
75%
75% of sessions included a verify step
correction rate
50%
50% of sessions involved correcting or challenging the AI
outcome-linked
100%
100% of sessions tied to a shipped outcome
ai tools used
claude code
5 · 63%
claude.ai
2 · 25%
chatgpt
1 · 13%
most common workflow shapes
  • analyze → decide×2
  • draft → analyze → decide → plan×1
  • reframe → analyze → communicate×1
  • research → analyze → decide → plan×1
  • analyze → debug → decide → plan → automate → review×1
interest fingerprint · 8 of 8 public
Auth systemstopicHigh-end audiopersonalTalent EngineeringtopicSports & cultural interestspersonalAI and frontier technology startupssectorFootball businesssectorSports managementprofessionalMulti-sector business experiencepersonal

the public part of inan's 8-tag active fingerprint. the rest stay private.

traits · from approved signals
privacy-risk intuition at the product design stage2layered consent architecture thinking2monetization-feature alignment2willingness to flag own uncertainty as a design signal2strategic reframing2stakeholder pressure navigation2narrative-first decision framing2product self-referentiality2category design under constraint2buyer redefinition2
approved work signals
inan kocatepe·observed · oauth·real work·confidence 78/100·approved

Inan identified k-anonymity risk in opt-in intersection discovery before spec was locked

Inan was designing the hiring-discovery feature for Ship 3 and caught — mid-sentence — that a simple opt-in toggle wasn't safe enough for rare intersections. The AI expanded this into a four-state per-tag consent model with a k-anonymity floor (k>=10), and Inan's framing directly shaped the monetization wedge: hiring-mode as the Pro-tier paywall. The session moved from product idea to durable, privacy-sound spec in a single exchange.

what this shows
privacy-risk intuition at the product design stagelayered consent architecture thinkingmonetization-feature alignmentwillingness to flag own uncertainty as a design signal
#privacy#product-design#consent-architecture#hiring#monetization
inan kocatepe·observed · oauth·real work·confidence 52/100·approved

Inan reframed Fenerbahçe wage cuts as a championship wage structure

Inan used AI to reframe a politically sensitive cost-cutting problem — Fenerbahçe's wage discipline — as a tiered championship structure rather than austerity, deflecting fan backlash. He also explicitly tested whether Granular's own framework (cost-cutting as a signal category) applied to the scenario, showing product-level self-awareness. The session demonstrates an instinct for narrative-first problem solving under stakeholder pressure.

what this shows
strategic reframingstakeholder pressure navigationnarrative-first decision framingproduct self-referentiality
#reframing#cost-cutting#stakeholder communication#product thinking#sports strategy
inan kocatepe·observed · oauth·real work·confidence 82/100·approved

Inan reframed Granular's category from AI work intelligence to AI-era credential layer for Talent Engineering

Inan fed the AI a dense, specific essay on how frontier companies (Cursor, Cognition, xAI, Stripe) actually hire, then demanded a sharp category reframe and concrete build/pricing/GTM direction — explicitly rejecting vague positioning. The session produced a full strategic pivot: new buyer (Head of Talent + VC Talent Partners, not CFO/HR), new price architecture ($50K–150K enterprise, $30K VC seat), and a prioritized 8-9 week build order with explicit 'do NOT build' constraints. The prompt itself showed disciplined problem framing — Inan named the category he feared settling for and asked the AI to beat it.

what this shows
category design under constraintbuyer redefinitionGTM sequencing disciplinestrategic elimination (what NOT to build)competitive displacement framing
#category design#talent engineering#gtm strategy#positioning#product roadmap
inan kocatepe·observed · oauth·real work·confidence 82/100·approved

Inan stress-tested a PLG theory and committed to a single warm pilot

Inan posed a binary product-direction decision under real resource constraints (solo founder, two-week horizon) and explicitly asked the AI to push back on the weaker conversion theory — not just pick one. The AI's rejection of B is grounded in a precise structural argument (self-consumed artifact, no team-pull mechanism, wrong-persona data), and Inan's prompt design forced that level of scrutiny rather than a hedged 'it depends' answer. The session closed with a concrete two-week action plan tied to a willingness-to-pay test, linking the decision directly to a falsifiable outcome.

what this shows
forced trade-off framingriskiest assumption prioritizationPLG mechanism scrutinypilot scoping over abstract market commitmentmonetization path discipline
#product strategy#go-to-market#plg#solo founder#decision-making
inan kocatepe·self-reported·real work·confidence 72/100·approved

Inan interrogated the integrity gap between self-reported and observed-user-work tokens in Granular's OAuth layer

Immediately after an OAuth handshake completed on a live deployment, Inan asked not what the feature does but what it actually proves — and explicitly named the axis where trust still breaks down (session sharing vs. seat ownership). This is an active architectural audit, not a curiosity question. The framing reveals he holds a mental model of the trust hierarchy (observed > self-reported) and is already stress-testing where it fails before it ships.

what this shows
trust-model reasoninggap identification under live conditionsarchitecture interrogationhonest constraint acknowledgment
#oauth#trust-model#integrity-layer#product-architecture#granular
inan kocatepe·self-reported·real work·confidence 81/100·approved

Inan audited and repaired Granular's auth and identity layer before shipping new features

Inan framed the session with a precise constraint — fix what exists, don't build new — and front-loaded an auth invariant check (Inan owns inan, Hugo cannot reach /manager) before touching any code. The session surfaced a non-obvious root cause: a legacy auto-inherit branch in /api/seats/claim that silently assigned founding-team identity to any new user who matched a pre-seeded slug, which is the kind of subtle identity bug that typically survives for weeks. Inan coordinated fixes across six surfaces (DB, two API routes, signup form, profile edit, manager header) and capped it with Resend SMTP wiring, including an honest caveat about sandbox sender spam risk.

what this shows
root-cause isolation before fixingpre-condition verification disciplineconstraint-scoped execution (no feature creep)multi-layer system thinking (DB → API → UI → email)honest risk flagging
#auth#debugging#identity#product-infra#email
inan kocatepe·self-reported·real work·confidence 78/100·approved

Inan audited Granular's auth and identity layer before touching anything else

Inan structured the session around a mandatory pre-condition — verify the core auth invariant (Inan owns inan, Hugo cannot reach /manager) — before authorizing any code changes. The AI then traced Hugo's confusing signup to a silent auto-inherit bug in /api/seats/claim, fixed it across six surfaces in one commit, and honestly flagged the remaining blocker (OAuth tokens) and the spam-risk caveat on the unverified sender domain. The prompt shows strong scope discipline: fix what's broken, no new features, and brand the emails.

what this shows
auth-first debugging disciplineroot-cause isolationscope enforcement under pressureinfrastructure layering awareness (SMTP vs. Supabase mailer)honest caveat flagging
#auth#debugging#identity#email-infrastructure#product-ops
inan kocatepe·self-reported·real work·confidence 78/100·approved

Inan differentiated Granular's wedge motion across two structurally different pilot buyers

Inan diagnosed that Monay and Prowin require fundamentally different pitch entry points — not just different messaging but different buying triggers and readiness conditions. He explicitly invited pushback on his two candidate wedges for Prowin, which surfaced a third framing ('training and consistency') the AI introduced and he did not pre-load. The session produced a concrete qualification gate: if the Prowin buyer can't immediately name what their most experienced person does differently, the pilot should be deferred.

what this shows
segmentation thinkingwedge differentiationpushback solicitationpilot sequencing discipline
#go-to-market#pilot strategy#segmentation#positioning