AGENTIC CAREER INTELLIGENCE · EIGHT ESSENTIAL AGENTS

Say “help me get this job.”
The platform does the rest.

Paste a job description and your resume. Lattice understands the role, reads your evidence, scores your hiring readiness from three independent reviewers, and interviews you out loud — then remembers what to fix next time.

SCALE-READY BY DESIGN

Built on a production-grade,
multi-agent stack.

Next.js
Next.js
React
React
Tailwind CSS
Tailwind CSS
tRPC
tRPC
Drizzle ORM
Drizzle ORM
ONE GOAL, EIGHT MOVES

From a job description to an offer-ready candidate

The Career Orchestrator turns your goal into a plan and runs independent work in parallel — concurrent agents, not a sequential checklist.

01Understand the jobThe Job Agent parses your pasted description into a structured profile.
02Read the resumeSections and structure preserved for citation, chunked and embedded locally.
03Talk it throughReal-time chat about your experience, grounded in your own file.
04Retrieve evidenceEvery answer backed by retrieved resume chunks, never invented.
05Evaluate readinessATS, recruiter, and hiring-manager reviewers score you concurrently.
06Resolve disagreementWhen reviewers diverge, one is re-run with the others’ reasoning — or the gap is shown, not hidden.
07Interview liveVoice session with Gemini Live or Pipecat, transcript captured in full.
08RememberScores update your candidate profile so the next round starts ahead.
WHY LATTICE

Most tools rewrite your resume.
This one builds a case for you.

01GROUNDED IN YOUR EVIDENCE

Claims, tied to the line that proves them

Every skill in your profile carries the sentence and section it came from. Anything unsupported gets flagged as missing evidence rather than quietly asserted.

Resume chunked, embedded and retrieved per question
Span-level attribution after generation
Missing evidence surfaced, not invented
02THREE REVIEWERS, ONE READINESS SCORE

ATS, recruiter and hiring manager — run in parallel

Three independent specialists score you concurrently on different dimensions. When they disagree, the outlier is re-run once with the others’ reasoning before anything is finalized.

Concurrent scoring, not sequential
Bounded critique/refine loop
Unresolved disagreement shown explicitly
03INTERVIEW SCORING, NOT A TEMPLATE

Every score comes from what you actually said

Live voice interviews are transcribed in full. Evaluation is derived from the transcript by a dedicated specialist — never a hardcoded rating.

Gemini Live or Pipecat (LiveKit), your choice
Pipecat barges in mid-sentence, like a real call
Learning roadmap generated per session
04MEMORY THAT COMPOUNDS

The next application starts ahead of this one

Confirmed weaknesses from each interview roll forward into your candidate profile — reviewable, and never written without an evaluation behind it.

Recurring weakness detection across sessions
Profile updates tied to real evaluations
One evidence-backed candidate profile
ASK YOUR RESUME

Every answer cites the line it came from

Your resume is chunked, embedded locally and retrieved per question, so the chat answers only from your own evidence.

Retrieval scoped to youChunks filtered by authenticated identity and resume version.
Answer streams firstCitation attribution attaches after — it never blocks the response.
Fail-soft citationsIf attribution is down you see “verification unavailable,” not an error.
CHATRESUME · SCOPED TO YOU
What evidence do I have for backend development?
GROUNDED ANSWER

Your resume shows backend experience through Django API development and PostgreSQL data pipelines.

EXPERIENCE · CONFIDENCE 0.94Built Django APIs for a production application.
PROJECTS · CONFIDENCE 0.88Built a data pipeline using Python and PostgreSQL.
EVIDENCE ATTACHED · 2 SPANS VERIFIED
◆ HIRING READINESS, EXPLAINED

Three reviewers, and the reason behind each score

ATS, Recruiter, and Hiring Manager agents score you concurrently, not in sequence. Nothing is scored without a stated reason — disagreements are shown, not averaged away. A dedicated Job Match node scores 11 weighted dimensions against this exact posting, alongside it.

81OVERALL READINESS
BACKEND ENGINEER · EXAMPLE CORP
EVIDENCE-BACKED
ATS PARSING92
RECRUITER APPEAL78
TECHNICAL FIT81
EVIDENCE QUALITY70
PRIORITY 01Attach a metric to the PostgreSQL pipelineEvidence quality is the lowest score; the claim exists but carries no number.
PRIORITY 02Move the API performance project above courseworkRecruiter appeal improves when relevant work leads the page.
FLAGGEDAWS appears in the job description with no supporting experienceLeft out of the tailored view rather than claimed.
TAILORED MATERIALS, NOTHING INVENTED

It scores the fit, drafts the letter, and finds what else is out there.

62/100 · CAPPED
Job match score11 dimensions, weighted not averaged
One missing required skill visibly caps the score
Semantic similarity from your real resume chunks
Cover letterSpecific to the company
Built from job profile + your evidence
Never sent without your review
ARBEITNOWREMOTIVEJOBICYREMOTEOK
1
2
3
Similar rolesAggregated from 4 job boards
Arbeitnow, Remotive, Jobicy, RemoteOK — deduped
Ranked by semantic similarity + skill overlap
Eight moments the workflow stops and waits for you
REPEATS EVERY APPLICATION
01Approve the agent plan
02Pick the target role
03Confirm low-confidence claims
04Export the cover letter
05Approve an outreach draft
06Attest a manual send
07Opt in to reminder emails
08Approve a Notion write
JOB DUMP · MOMENTUM · REFERRALS

Save everything, triage on your terms, never lose the thread

SAVED
APPLIED
STALE
Job Dump & Momentum AgentTwo views of the same saved roles, your call
Structured Kanban or a scribble corkboard view
Deterministic staleness/deadline nudges — no LLM, nothing hidden
PUBLIC-CONTEXT
PUBLIC-CONTEXT
Referral discoveryGitHub-sourced leads today, LinkedIn via Edges next
Every lead labeled "public-context," never "warm connection"
Editable outreach drafts — no send button exists, anywhere
Career memory graphPeople, companies and schools, evidence-linked
Backfilled from resume + saved jobs you already have
Every relationship click-through to its source
INTERVIEW & GROWTH

Live voice interviews, scored from the transcript

Gemini Live or Pipecat on the other side — interrupt mid-sentence and it responds naturally — a plan weighted to the role, and evaluation derived from what you actually said, not a template.

RESUME20%
TECHNICAL40%
BEHAVIORAL30%
COMPANY10%
◆ CAREER MEMORY

A candidate profile that improves with every application

Confirmed weaknesses roll forward. Reviewable, and never written without an evaluation behind it.

Skills and proficiency18 tracked
Approved achievements11 confirmed
Resume versions4 stored
Interview transcripts6 scored
Recurring weaknesses2 open
THE ASSISTANT

One launcher. A brief that reads your day, and an agent that orders dinner.

Every read is live. Your real signals, real restaurants, a real cart. Every write stops and asks you first.

+ - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - WAYS THE ASSISTANT SHOWS UP IN YOUR DAY+ - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + -
MORNING BRIEF
Good morning. Two deadlines
need you today.
READINESS78
DEADLINES2
HIGH MATCH3
01

Your morning brief

Reads yesterday from your own data — readiness moves, deadlines closing, roles worth a look — and writes the one paragraph that matters.

SWIGGY AGENT
Dinner under ₹400 tonight
search_dishes· live
browse_menu· live
Paneer Butter Masala
⭐ 4.4 · 32 min
₹289
02

Swiggy agent

Ask in plain language. It searches live restaurants and real menus through MCP, then proposes an order with actual prices.

ORDER JOURNEY
ALLVEG
₹289ADD
₹340ADD
₹180ADD
CART
₹809
03

Browse it yourself

Or skip the chat: real menus, a veg filter, a live Swiggy cart with the server’s own fees and taxes — kept in its own tab.

APPROVE, THEN PAY
APPROVEREJECT
Pay on deliveryCOD
Swiggy MoneyWALLET
SIMULATED₹809 paid
04

Approval, then actual order

The cart and the payment methods are genuinely yours. Placing the order is always simulated — no restaurant is ever sent anything.

BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·BRIEF·ORDER·APPROVE·
THE ONE THING IT WON’T DO

Why don’t we have auto-apply?

Because the second a machine can apply for you, you inherit everything it got wrong, at the speed it can send. A bad match, a hallucinated claim, a recruiter who now remembers your name for the wrong reason.

Approve this run
The agent has stopped here. Nothing happens until you answer.
1What it wants to do
Place a food order · 2 items · ₹809 · pay on delivery
2Your call
APPROVEREJECT+ reason
[ THE GATE ]

It stops, and asks

Every write pauses the graph on a real interrupt — not a confirm dialog bolted on afterwards. Approve it, or reject it with a reason the agent reads before it tries again. There are eight of these moments, and none of them is “submit application”.

REFERRAL AGENT

Draft ready — 3 leads, public context only. No warm intro is implied.

YOU
COPYno send button exists
I sent this myself — logged as manual
[ NO SEND BUTTON ]

Nothing leaves without you

Outreach is drafted, never dispatched: there is no send path in the codebase to begin with. You copy the message, send it from your own account, and mark it sent — so the record matches what actually happened.

Eight agents. One goal. Your evidence.