San Diego pilot · process development groups

Batchrite is the digital
batch record for process development.

Built for process development groups — cell therapy, mRNA / LNP, oligonucleotide, bioconjugate, AAV. Draft your process as a graph, capture every lot, and use AI to help you do it — so you never lose a parameter, or the reason a run drifted, even when the scientist who ran it has moved on.

Graph-basedprocesses as graphs, not step-lists AI assistantdrafts your process as a graph Audit trailevery change, who and when Role-based accesspermissions scoped per project Pre-GLPfit for PD, not commercial Process memorysurvives scientist turnover Your data, your wayopen formats, no lock-in
app.batchrite / protocols / lnp-formulation · mrna-construct-a4 · v3
+ node save designer
pd scientist
process lead
analytical
head of pd
START begin run
UNIT-OP thaw mRNA-A4 lot mRNA-2604-12
UNIT-OP prep lipid mix SM-102 · DSPC · chol
UNIT-OP LNP formulation N/P 6 · 12 mL/min · running
IPC size + PDI DLS · 60–100 nm
REVIEW process lead e-sig · review & lock
RECORD lock batch EBR → process memory
7 nodes · 4 swimlanes · SvelteFlow · v3 auto-saved 12s ago
PD in 2026

The hardest molecules in the industry, recorded on the worst tools.

You’re running CAR-T, LNP, AOC, or AAV development with three to fifteen PD scientists. Every run is novel. Every lot matters. And the institutional memory of how each process actually behaves is sitting on clipboards, in SharePoint folders, and in the head of the scientist who left last month. Batchrite is the fix.

01 / problem
.pdf

Paper batch records still rule the PD suite.

Runs are logged on clipboards, transcribed at 18:00, and re-keyed into a SharePoint spreadsheet at 22:00. By the time CMC asks why lot 6 looked different, the answer is gone.

02 / problem
.eln

ELNs weren’t built for processes-as-graphs.

Real PD processes branch, loop, and parallelize. Linear step-lists flatten the truth, then patch the model with spreadsheets and tribal knowledge that walks out the door with every new hire.

03 / problem
.ai?

AI hasn’t arrived in the PD suite yet.

Generative models can draft a 200-page legal brief, but the PD scientist is still pasting prompts into a chat window and copying answers back into a paper form headed for a three-ring binder.

Built for process development

Six features that make Batchrite different.

From the graph-based process designer to AI-assisted authoring, vision capture, and paper-record import, every primitive is purpose-built for how process development groups actually run novel processes today.

01 · Process designer

A visual, graph-based editor — swimlanes for every PD role.

Drag-and-drop unit operations onto a canvas. Assign roles, bind equipment slots, branch logic, parametrize every step. Built on SvelteFlow so the canvas feels as fast as Figma.

Processes are templates. Each run snapshots them at execution time and tracks every deviation against the source. Refactor a template and your historical runs stay intact — the process memory survives turnover.

  • Swimlanes per role (PD scientist, process lead, analytical, head of PD)
  • Equipment-slot binding at design time
  • Branches, loops, parallel arms
  • Templates → snapshots → deviations
  • Parametrized everything · units enforced
Designer · node detail · LNP formulation
v3 · auto-saved
UNIT-OP thaw mRNA-A4 lot 2604-12
UNIT-OP · 6/11 LNP mix N/P 6 · 12 mL/min
IPC size + PDI DLS
selected node
LNP mix
UNIT-OP · process lead lane
N/P ratio
6.0 ± 0.2
flow rate
12.0 mL/min
equipment slot
lipid mixer · bench-A
role
@process-lead
02 · AI process authoring

Describe what you need. The agent drafts the graph.

Type a prompt. Get unit operations, parameters, swimlanes, and IPCs. Iterate in chat until it’s right, then commit to a new process version. The whole graph stays editable in the designer.

  • Generates swimlanes & role assignments
  • Parametrizes from your prompt
  • Iterate in chat · diff each pass
  • Validates against your existing templates before commit
Assistant · process draft
ready
you
Draft an LNP formulation for mRNA-A4 at 100 mg scale. SM-102 / DSPC / chol / DMG-PEG. N/P 6. Add IPCs for size, PDI, encapsulation.
assistant
Drafted a 14-step graph across 4 swimlanes. Pulled mixing parameters from your last mRNA-A2 process and added IPCs at Ribogreen, DLS, and TFF. Three deviations from your reference LNP template flagged for review.
Open in designer → 3 deviations View diff
03 · Literature retrieval

Pull processes from OpenWetWare — with a human-in-the-loop gate.

Ask the agent to find a process from the public literature. It searches OpenWetWare, returns candidates with full attribution under CC BY-SA 3.0, and — only after you approve — drafts a Batchrite process from the source.

No hallucinated steps slip into your library. Every imported process carries a citation trail.

  • Search OpenWetWare & public protocol libraries
  • Attribution + CC BY-SA 3.0 preserved
  • Human approval required before import
  • Source diff at every step of the draft
Literature · OpenWetWare search
3 candidates
🔎 LNP formulation SM-102 DSPC mRNA
LNP formulation · SM-102 / DSPC / chol
match · 91%
OpenWetWare · contributed by Anderson Lab, UCSF · last edited 2024-11
14 steps 3 IPCs CC BY-SA 3.0
Approve & import → Preview source
⚠ HUMAN-IN-LOOP
Nothing imported until you approve. We won’t draft hallucinated steps.
04 · Process memory

Every run, every parameter, every deviation — remembered.

Batchrite gives PD groups what paper records and SharePoint can’t: a single, queryable history of every run you’ve done. Lightweight review-and-lock workflow (PD scientist → process lead → head of PD) with hand-drawn e-signatures. Immutable audit trail. Reason-for-change on edits. Built for PD — not bolted on as commercial GMP overhead.

  • Scientist → process lead → head of PD review flow
  • Hand-drawn digital signatures (initials & full)
  • Immutable audit trail on every mutation
  • Reason-for-change capture on locked records
  • Search across every lot, every parameter, every observation
  • Survives scientist turnover · the process remembers
Review & lock · BR-2026-0184 · head of PD
locked at signature
Scientist
Process Lead
Head of PD
Locked
SIGN HERE · J. PARK · HEAD OF PD
reason for change none · final review
05 · Paper-record import

Got a stack of paper records? Drop them in.

Upload a filled-in paper batch record — PDF, DOCX, or image. The AI extracts step status, timestamps, parameter values, signatures, and notes; maps fields to your digital run; and backfills the execution record — so your historical lots become searchable process memory.

  • PDF · DOCX · image (PNG / JPEG)
  • Extracts steps, timestamps, parameters
  • Signature & deviation detection
  • Manual override on every mapping
  • Backfills historical runs at scale
Import · batch-record-2025-09-12.pdf
extracting · 87%
BATCH RECORD · BR-2025-0911
Date: 12 Sep 2025
Operator: J. Park
Step 1 · prep · 10:14
Step 2 · mix · 10:46
Step 3 · sample · 12:11
OD₆₀₀: 0.82
notes: foam at 14m, added 2mL af
Head of PD sig: J. Park
extracted
date2025-09-12 ✓
operatorj.park ✓
step 1 · prep10:14 ✓
step 2 · mix10:46 ✓
step 3 · sample12:11 ✓
OD₆₀₀0.82 ✓
deviation notereview →
06 · Vision capture

Snap a photo. The agent reads it.

Tap the camera during a run to capture a balance display, gauge, plate, gel, or handwritten log. The vision model parses the image, lifts numeric readings, lot codes, and observations, and proposes structured fields against the current step. You approve before anything lands in the batch record.

  • Reads digital displays, analog gauges, labels & barcodes
  • OCR for handwritten notes & printed log sheets
  • Auto-binds readings to the step’s expected parameters
  • Per-field confidence with manual override
  • Original photo retained · audit-trail linked
Capture · step 6 · mix · T+14:32
analyzed
CAMERA · REAR ● REC
0.821 g LAB-04 cal · 9/01
tap to snap · auto-detect on
vision read 98% confidence
sample mass0.821 g ✓
instrumentbal · LAB-04 ✓
calibration2025-09-01 ✓
lot labelmRNA-A4-09-12?
timestampT+14:32
Everything else, baked in

The full surface area.

The seven headline features get the spotlight. The rest of the product is everything you’d expect from a PD-focused batch-record system — without the four-month implementation cycle that commercial MES vendors quote you.

Run execution
  • Parameter & equipment overrides without changing the original process
  • Per-step notes & media (photo, video)
  • Run history timeline with branch visualization
  • Lot suggestions & COA attachment
Documents
  • Auto-generated batch records & protocols from your templates
  • PDF + DOCX output, branded to your team
  • AI document refinement · filled form → reusable template
  • Full-text library search
Analytics
  • Pass/fail/deviation rates per process
  • Step-duration variance across runs
  • Lot-to-lot variance views per modality
  • Version-history diff viewer
Equipment & suites
  • Serialized inventory tracking (lots, columns, instruments)
  • Calibration schedule & alerts
  • Equipment-slot binding at design time
  • Multi-suite management with role scoping
Collaboration
  • Organizations & teams
  • Fine-grained roles & permissions
  • Email invites scoped to a project
  • Slack & email notifications
AI vision
  • Per-image & batch image analysis
  • Extract numeric readings from gauges & displays
  • Anomaly detection across run images
  • Visual confirmation of expected outcomes
Data ownership

Your data is yours. Always exportable.

No proprietary file formats. No “contact sales” to get an export. No retention games when you cancel. Every protocol, run, batch record, signature, photo, and audit-log entry is yours — available in open formats, on demand, for as long as you have an account.

01

One-click full export, any time.

Pull the entire project — protocols, runs, batch records, signatures, attachments — as a single archive. Or per-run as Excel, CSV, or PDF.

02

Open formats · no lock-in.

Excel, CSV, PDF, and image files — the formats your team already uses. Walk in with paper, walk out with data you can actually open and read — and walk away whenever you want.

03

Connect it to the systems you already use.

Send every run to your team’s own data systems automatically, and pull anything back out on demand. Get a notification on every sign-off. Your systems stay your systems.

04

Never used to train shared models.

Your data is kept separate from every other customer’s. AI features work over your data, for your queries, only. Disable them entirely if your compliance posture requires it — the product still works.

Export · project · acme-bio
12.4 GB · 847 runs
scope
format
Excel CSV PDF Images
typically ready in < 4 min · secure link · resumable
OUR PLEDGE

If you ever decide to leave, we’ll get you and your data out the door cleanly. 90-day post-cancellation access. Full archive on request. No exit fees, no negotiation, no friction.

Read the export terms →
The San Diego pilot

Join the process development cohort.

Batchrite’s first cohort is 15 PD groups in the San Diego biotech cluster — cell & gene, mRNA, oligo, bioconjugate, AAV. There’s no self-serve signup. Every partner starts with a 30-minute conversation so we understand your modality and bench before you ever touch the product.

Every partner starts with a call. Pilot runs through Q4 2026. Coming to Cell & Gene Meeting on the Mesa? We’ll be there — email hello@batchrite.com and we’ll find 20 minutes between sessions.
Trust & data · built for PD, not GMP

Fit for purpose: pre-Phase 1 PD, not commercial manufacturing.

Batchrite is built for the work you’re doing before the GMP shadow falls — PD, late-stage research, IND-enabling development. We don’t claim 21 CFR Part 11 compliance and we don’t pretend to be a GxP MES. Here’s what we do ship — the traceability and process-memory primitives a PD group actually needs — and what we don’t pretend to.

Shipped

Lightweight review & lock

Scientist → process lead → head of PD. Each step gates the next; post-lock edits require a fresh signature and reason. Right-sized for PD — not GMP overhead.

Shipped

Hand-drawn digital signatures

Initials and full signature, bound to every consequential review event. Stored as immutable artifacts on the audit trail.

Shipped

Immutable audit trail

Every mutation logged with actor, timestamp, before/after. Append-only. Tamper-evident hashes per entry — so process memory survives turnover and CMC handoff.

Shipped

Reason-for-change capture

Post-lock edits require a typed justification and a fresh signature. Locked records can’t be silently mutated.

Shipped

Lot-traceable run history

Every lot, reagent, instrument, and operator threaded through a queryable history. When CMC asks why lot 6 looked different, you have the answer in seconds.

Shipped

Full data export · no gatekeeping

One-click archive of the entire project in open formats (Excel, CSV, PDF). Per-run exports, or connect Batchrite to your own data systems. Your data, on demand.

Shipped

Customer data isolation

Each customer’s data is kept logically separate. Your data is never used to train shared models. A dedicated, single-customer deployment is available for design partners.

In progress

SSO & SAML / SCIM

SAML SSO and SCIM provisioning with Okta, Entra, and Google Workspace. Targeting Q3 2026 with design partners.

In progress

Encryption with your own keys

Data is encrypted at rest today. The option to bring your own encryption keys (AWS, GCP) is rolling out to design partners through Q3.

Not in scope

21 CFR Part 11 / GxP claims

We don’t claim Part 11 compliance, GLP validation, or commercial-GMP fitness — and we’ll tell your QA group as much in writing. Plan to move to a validated MES (Tulip, MasterControl, Veeva) once your process leaves PD. Batchrite is built to hand off cleanly when that day comes.

Shipped · in production for the SD pilot In progress · with design partners Not in scope · out of bounds by design
Coming soon

Three things we’re building next.

Founding partners vote on what jumps the queue. Here’s what’s landing first.

FUT-0005 Q3 2026

Process intelligence & DoE analysis.

Ask the chat agent: “Analyze my last 12 LNP runs and tell me what’s driving % encapsulation.”

batchrite assistant

N/P ratio (p = 0.003) and aqueous flow rate (p = 0.01) are your significant factors. Encapsulation peaks at N/P 6.2 / 12.5 mL/min. Want me to draft a new process version with these parameters?

→ replaces JMP / Design-Expert for the 80% case

FUT-0011 Q4 2026

CDMO export & tech-transfer packs.

One click turns a locked process into a tech-transfer package — mapped to your CDMO’s format, or your own custom SOP / batch-record template.

+ CDMO transfer format + Custom SOP template + Batch-record template + Parameter + spec tables

→ no more rebuilding the record by hand for transfer

FUT-0016 Q4 2026

MCP server for your local agents.

Connect Batchrite to the AI agents you already run. Query runs, draft processes, and log results — without ever opening the UI.

your local agent · via MCP

Logged OD₆₀₀ = 0.82 to run BR-2026-0184 and flagged the 14-min foam event. Drafted next week’s LNP run from the v3 template — ready for your review.

→ the bench talks to Batchrite, no tab-switching

FAQ

Pilot questions, answered.

Working with an early product is a different kind of decision than buying a mature one. Here’s what process development groups typically ask before signing on.

Are you GLP / 21 CFR Part 11 compliant?

No — and on purpose. Batchrite is built for process development work, which is pre-GMP and pre-Part 11 by design. We don’t make Part 11 compliance claims and we don’t pretend the product is fit for commercial GMP manufacturing. We’ll tell your QA group as much in writing.

What we do ship: a lightweight scientist → process lead → head of PD review-and-lock flow, hand-drawn e-signatures, an immutable audit trail, and reason-for-change capture. That’s the right floor for PD — enough traceability to defend a process to CMC, without the GxP overhead that would slow you down before you have a clinical candidate.

When your process moves into GMP, we’ll hand off cleanly to a validated MES (Tulip, MasterControl, Veeva) with a full export of the development history.

How is this different from Benchling, eLabNext, or LabKey?

Those are ELNs and LIMSs. They’re great at structured inventory, sample tracking, and the discovery side of the house. They were not built to be a batch record system for a complex, multi-step PD process — which is why every PD group we’ve talked to runs Benchling alongside a stack of paper or SharePoint spreadsheets.

Batchrite is the digital batch record layer for that PD process — graph-based, modality-aware templates, AI-assisted authoring. It plugs in next to your ELN; it doesn’t replace it. Several pilot members run both.

We’re a 30-person Series B in San Diego. Is this for us?

That’s exactly who this is for. The pilot is sized for cell & gene, mRNA / LNP, oligo, bioconjugate, and AAV PD groups with 3–15 active scientists, pre-Phase 1 or Phase 1, currently on paper batch records or SharePoint / Excel hybrids. Series A or B, 20–80 employees is the sweet spot.

If you’re still 3 people in an incubator bay we’ll still talk — we just won’t pretend you need the whole product yet.

What happens to my data when the pilot ends?

Your data stays your data — full stop. You can export the entire project as a single archive (with all your files and attachments) at any point, or stream it continuously to your own data systems. No “contact sales” gate, no proprietary format, no exit fees. Done from settings in two clicks.

If you cancel, we keep your project accessible for 90 days so you can pull anything you missed. If you decide not to continue at GA, we’ll give you a 60-day grace period to migrate on top of that.

How does the AI authoring work? Where does the data go?

The assistant works only over your own data — your processes, templates, run history. Your proprietary information is never used to train shared models, and is only ever read to answer your own questions.

For the OpenWetWare integration, the search runs against the public corpus and returns candidates for your approval. We will not draft Batchrite processes from those candidates until you click Approve & import.

You can disable the AI features entirely if your security posture requires it. The product works without them.

How much will it cost at general availability?

We’ll publish GA pricing 30 days before the pilot ends. Expect it to be priced for Series A/B PD groups — not for commercial manufacturing.

Pilot cohort members get grandfathered rates locked for 24 months after GA — below whatever public pricing turns out to be.

Why aren’t you SOC 2 certified yet?

Because we’ve spent the first year of the company on the product, not the badge. SOC 2 Type II is on the roadmap for late 2026 — sequenced after GA so the controls we attest to are the ones actually in production.

In the meantime, we’ll walk your security team through our architecture and answer any questionnaire. The underlying controls (data isolation between customers, encryption at rest, an immutable audit trail, and role-based permissions) are already shipped.

How do you handle data coming off our instruments today?

Honestly: today, mostly by hand. PD scientists log readings and observations directly into Batchrite at the bench, or upload exports (CSVs, PDFs, photos of an instrument display) and let the AI structure them. Paper-record import handles the long tail of historical lots.

We don’t ship native instrument integrations yet — it’s the most common cohort request and we’d rather wait until we know which two or three instruments our pilot members actually live on, then build those adapters properly. You can connect Batchrite to your own systems for anything you want to wire up yourself in the meantime.

San Diego pilot · 8 design-partner spots remaining

Ready to digitize your process memory?

Book a 30-minute call to start. If you’re inside the 30-minute SD drive zone we’ll come to you with two coffees. Otherwise, we’ll meet over video.