Skip to content

R&D and product development: find past trials before you run another

The R&D agent reads your trial briefs, trial logs, lab-dip cards and sample register. Ask what an earlier trial used and concluded, or which samples wait on which buyer. Buyer emails go out only if an admin approves.

Works from: Trial brief, Trial log, Lab-dip card, Sample request

Where a development desk loses trials, samples and its GOTS or GRS claim

The trail is in records the desk already keeps: the trial brief, the trial log, the lab-dip card, the sample register, the TC that lists the yarn lot. What it costs is machine time, unpaid samples and developments that never become orders.

  • Samples go to the buyer, and the trail goes cold

    The desk sends a development to the buyer with its tech sheet and waits. Feedback comes by email weeks later, or never. Which developments sit with which buyer, and since when, is in a register nobody has updated. Better Buying’s 2018 report names one possible cause of low adoption: buyers taking a supplier’s samples to cheaper suppliers to produce. Against that risk, the playbook tags confidential trials buyer-only.

    61.4%

    Share of supplier ratings in which the buyer did not pay for samples, in Better Buying’s Q2 2018 ratings cycle (363 ratings) of apparel, footwear and household-textile buyers.

    Source: Better Buying, Better Buying Index Report, Fall 2018: Purchasing Practices Performance in Apparel, Footwear, and Household Textile Supply Chains (opens in a new tab), 2018.

  • A sustainable-fibre trial with the wrong paperwork behind it

    An organic-cotton or recycled-polyester trial runs well, but no transaction certificate on file lists the yarn lot, the one on file lists another lot, or the supplier’s scope certificate does not cover the product. The trial cannot carry the GOTS or GRS claim the buyer wants, and it runs again on a verified lot. The playbook calls chain-of-custody paperwork the trap and says to verify it at every trial.

    20 bodies

    Bodies that issue OCS organic-cotton certificates. Textile Exchange says chain of custody between facilities runs on paper and PDF certificates, and that some facilities keep finding ways to substitute certified fibre.

    Source: Textile Exchange, How organic cotton certification works (opens in a new tab), 2023.

  • The same trial, run twice because the first one was lost

    A buyer asks for a recycled-cotton jersey or a peached twill with a cooler hand. Someone remembers a similar trial, but its log sits in one technician’s Excel, the lab-dip card in a folder and the swatch in a box. Nobody finds it in time, so the trial runs again on knitting, loom or dye-batch time the plant could have sold. The playbook puts it plainly: trials repeat themselves silently.

  • Trial costs that nobody adds up

    Yarn, machine time and chemicals for a trial get booked to production, or to nobody. The success criteria were never measurable, so the trial runs a third and a fourth cycle. At the monthly review, nobody can say what a development cost against its budget, and the owner pays for development without seeing the bill.

Past trials, sample approvals and TCs, answered from your records

It works from the trial records, registers and certificates uploaded to R&D, and its connected mailbox and sheets. Each answer names the trial log, card or certificate it used. A follow-up email or a register update waits for an admin to approve or decline it.

  • Past trials, found by fibre, blend, finish or buyer

    Reads your documents

    Upload trial briefs, trial logs, lab-dip cards and test reports to R&D. Ask about a concept and the agent lists the trials on record that match: what the brief asked for, the conditions recorded, the result and the conclusion, with the file each one came from. If hand-feel or GSM was never recorded, it says so. Whether an old trial is close enough is your call.

    Ask“Have we run a peached cotton twill with a cool hand before? Show each trial, its conditions and what it concluded.”

  • Hit rate and trial cost, counted from your rows

    Answers with sources

    Upload the sample register or trial cost sheet as Excel or CSV, with buyer, season and status columns; in Excel it counts from the first sheet, up to 5,000 rows. Each count gives one number and its rows: developments sent to a buyer in a season, those approved, or one trial’s total cost. A Google Sheet is read and cited, not counted. The playbook’s targets: approval of 40% or more, cost within 10% of budget.

    Ask“Of the developments we sent this buyer last season, how many were approved? Show the rows you counted.”

  • The TC that lists a trial’s yarn lot

    Reads your documents

    Upload scope certificates and transaction certificates (TCs) with the trial logs. Ask about a GOTS, OCS or GRS trial. The agent finds the lot it used and quotes the TC that lists that lot: its number, the product, the shipment and the dates. If no TC on file lists the lot, it says so. It reads text PDFs; a scanned certificate has no text to read. It does not check a certificate with its issuer.

    Ask“Which yarn lot did the GRS polyester trial use, and which transaction certificate on file lists that lot?”

  • A development summary, brief beside result

    Creates files

    Ask for a summary of a concept and the agent builds it as a document: what the brief asked for, each trial’s recorded result against each success criterion, the buyer’s comments, and what remains open. A criterion with no recorded result is marked as missing, never filled in. Ask it to rework one section, such as the open points, and only that part changes. Every earlier version is kept and attributed.

    Ask“Prepare a development summary for the stretch denim concept: what was tried, what worked, and what is still open.”

  • Buyer follow-ups and register updates, drafted for approval

    Acts in your tools

    From the sample register, the agent lists developments still waiting on the buyer, oldest date sent first, with the last comment on record. It prepares the follow-up email to the buyer’s developer, to go through Gmail or Outlook, or the status change for that row in your Google Sheets register. Each draft waits for the R&D head or another admin to approve or decline it, and only an approved one is sent or saved.

    Ask“Which samples to this buyer have no decision yet? Draft a follow-up to their developer that lists them.”

A repeat request, a trade show and a go/no-go review

  • A buyer asks for something you may have tried

    Fabric developer

    A buyer’s developer asks for a recycled-cotton jersey, and someone recalls a trial like it last year.

    1. Asks

      “Have we run a recycled-cotton jersey trial before? Show what it concluded, the blend and conditions it used, and its lab-dip cards.”

    2. The agent

      Searches the trial logs, briefs and lab-dip cards uploaded to R&D. Lists each recycled-cotton trial it finds with its date, buyer, blend, yarn count, machine gauge, stitch length and GSM as recorded, and its result against the brief. Adds any note on neps or hand-feel. Cites each file and says which fields were never recorded. If a trial recorded high neps, it quotes the playbook’s note on why: recycled fibre is shorter than virgin cotton.

    3. Result

      The developer starts from the earlier trial’s record instead of a blank brief, and decides whether a new trial is needed and what blend it uses.

  • Buyer-only samples kept off the trade-show stand

    R&D executive

    A week before a trade show, Sales & Marketing asks R&D for swatches of this season’s developments.

    1. Asks

      “Which samples in the register are tagged buyer-only? Draft a WhatsApp message to the sample-room in-charge listing them, so they stay out of the trade-show set.”

    2. The agent

      Reads the sample register in your connected Google Sheet and lists each sample tagged buyer-only, with its sample number, buyer, date and storage location. Quotes the playbook: a confidential trial is not shown on a trade-show stand without the buyer’s approval. Drafts a WhatsApp Business message to the sample-room in-charge with the list.

    3. Approval

      The message waits for the R&D head, as department admin, to approve or decline. Only an approved message goes out, and the decision stays in the audit trail.

    4. Result

      A list of the buyer-only samples the register records, each row cited, and a message that is sent only if approved. Holding the swatches back stays with the sample room.

  • A trial that has waited weeks for a decision

    R&D head

    A completed trial has waited weeks for a go or no-go, and the monthly development review is tomorrow.

    1. Asks

      “Prepare a development summary for the lyocell-blend shirting concept: what the brief asked for, what each trial recorded, and what is still open.”

    2. The agent

      Reads the brief, the trial logs, the test reports and the buyer’s comments uploaded to R&D. Sets each success criterion beside the recorded result and marks the ones with no result. Notes how many cycles have run, and quotes the playbook: after two cycles that do not converge, re-align the brief with merchandising instead of running a third. Builds the summary as a document.

    3. Result

      The review chooses to iterate, descope or pivot from one document, with every result cited and every gap named.

What unadopted samples cost, and what R&D should measure

These are published industry figures on hit rates, sample payment and sustainable-fibre sourcing, each cited at the source. None of them measures Mavick. Before a pilot, count your own approval rate, trial cost and lead time, then count them again at its end.

  • Knowing which developments each buyer adopts

    Nearly 3 in 10

    Suppliers whose buyer adopted 39% or less of the styles they developed, in Better Buying’s Q2 2018 ratings cycle (363 ratings). The playbook’s target for approved trials is 40% or more.

    How the agent helps

    The agent does not present or sell. When you ask, it lists the developments waiting on each buyer, oldest first, with the last comment on record, and counts how many were sent and how many approved, from the rows of the sample register you upload, and gives the approval rate. Any follow-up it drafts goes to the buyer only if an admin approves it.

    Track: Trials to buyer-approved sample rate, by buyer and season

    Source: Better Buying, Better Buying Index Report, Fall 2018: Purchasing Practices Performance in Apparel, Footwear, and Household Textile Supply Chains (opens in a new tab), 2018.

  • Trial spend you can see against its budget

    27.3%

    Share of supplier ratings in which the buyer paid for samples and paid on time, in Better Buying’s Q2 2018 ratings cycle (363 ratings). In the rest, the supplier waited for payment or carried the cost.

    How the agent helps

    Upload the trial cost sheet with a trial column, and the agent sums one trial’s recorded cost rows, yarn, machine time and chemicals, and quotes the signed budget on file beside it. It never makes up a cost: a trial with no cost rows shows as not recorded. Costing a development for the buyer stays with Costing & Pricing.

    Track: Trial cost against trial budget

    Source: Better Buying, Better Buying Index Report, Fall 2018: Purchasing Practices Performance in Apparel, Footwear, and Household Textile Supply Chains (opens in a new tab), 2018.

  • Sustainable-fibre trials buyers ask for, with the paperwork on file

    76%

    Share of polyester from recycled sources in 2025 among the 384 brands and retailers reporting to Textile Exchange’s Materials Benchmark, against 12% of global polyester production.

    How the agent helps

    When you ask, the agent reads what each organic or recycled-fibre trial recorded, from processing behaviour to test results, and whether a TC on file lists its yarn lot, quoting the certificate. It does not verify certificates or decide a claim; that stays with Compliance & Sustainability.

    Track: Sustainable-fibre trials per quarter, with each lot on a TC on file

    Source: Textile Exchange, Textile Exchange’s Materials Benchmark shows promise of progress among reporting companies, while broader trends in material production remain concerning (opens in a new tab), 2026.

What to baseline before a pilot

Measure each one before the pilot starts, then again at its end.

R&D & Product Development: the KPIs to baseline before a pilot, and why each matters
KPI to baselineWhy it matters
Trials to buyer-approved sample rateThe playbook reports it monthly, with a target of 40% or more. Counted per buyer, it shows where development effort turns into orders and where it does not.
Trial cost against trial budgetThe playbook allows ±10%. Over 25% reads as a scoping problem, and past 150% of budget the trial goes to the R&D head and the CFO to pause or re-scope.
Time from brief to first sampleHow long a buyer waits for a first sample. The playbook sets 4 weeks or less for standard work and 8 weeks for novel work. Re-running an old trial adds to it.
Iterations per trial before a go or no-goThe playbook allows two cycles. A trial still not converging after two goes back to merchandising to fix the brief, not to a third run.
Sustainable-fibre trials per quarterThe playbook tracks it against buyer demand. Count beside it the trials whose yarn lot is listed on a TC on file, since an uncertified lot cannot support the claim.

What the agent works from

Records it reads

  • Trial brief
  • Trial log
  • Lab-dip card
  • Sample request
  • Sample register
  • Spec card
  • Test report
  • Transaction certificate (TC)

In your ERP

From the factory ERP it reads Odoo sales order and invoice headers (number, customer, total, status, date) and SAP S/4HANA sales orders. It reads no fabric or article lines, so it cannot show which development reached bulk, and it writes nothing there. Trial briefs, logs, lab-dip cards, TCs and the sample register come from R&D’s files and sheets.

One ERP connection serves the whole factory: SAP Business One, SAP S/4HANA, Odoo, Oracle NetSuite, Microsoft Dynamics 365, WFX Cloud ERP or Datatex NOW. This agent reads only the records named above; we check yours on the call.

In your tools

Reads buyer comments in the R&D mailbox in Gmail or Outlook, and finds a sample’s row in your Google Sheets register, when you ask. Drafts a follow-up to the buyer’s developer, a status change in the register, or a Slack or WhatsApp Business message to the sample room. Each one runs only after an admin approves it; a declined draft goes nowhere.

Tools named today: Gmail, Outlook, Google Sheets, Slack and WhatsApp Business.

Never made up

  • Trial recipes, blends or process parameters
  • Test results or performance claims
  • Certification status for a fibre or finish
  • Costs or yields for a development

If it is not in your connected data, the agent says so.

Belongs to another department

What the R&D playbook covers

  • SOP: from trial brief to pilot run

    The brief, from Merchandising or Sales, names the target buyer, the specs (fibre, blend, GSM, hand-feel, finish) and the success criteria. The trial is costed, with budget sign-off above a set threshold. It runs on one or two looms, one dye batch or one finish batch, then is tested against those criteria and documented.

  • Decision rule: a trial misses a success criterion

    Three options: iterate once more if the cause is known, descope by agreeing new criteria with the buyer, or pivot to a different approach. No more than two blind cycles without re-aligning with merchandising. The playbook’s reading: if a trial is not converging, the brief is wrong, not the trial.

  • KPI: time from brief to first sample

    Four weeks or less for a standard development, eight or less for a novel one. Each trial is reported with its log, its result and a spec card, and each month the playbook reports the trial-to-approval rate and the sustainable-fibre pipeline beside it.

  • Edge case: a certified supplier ships an uncertified lot

    If a GOTS supplier ships a non-GOTS lot by mistake, the trial cannot support a GOTS claim. The playbook says to document it and redo the trial on a lot with a verified chain of custody, and to check the supplier’s chain at every trial rather than assume it.

Written as general textile practice and reviewed by four independent AI models. Not yet signed off by a human subject-matter expert. Upload your own SOPs and standards, and every answer shows which source it used.

Ten questions it ships with

The template’s starter questions. Your team can edit them, or ask in its own words.

  • What a trial brief specified

    What did the brief for this trial specify, and what did the trial actually produce? Show both.

  • Lab dips for a concept

    Which lab dips or hand-feel developments exist for this concept, and what did each record?

  • Has a similar trial been run

    Have we run a similar trial before? Show what it concluded and what conditions it used.

  • Sustainable fibre trial results

    What did our trials with this sustainable fibre record — performance, processing behaviour, and any issues?

  • Developments pending by buyer

    Which developments are pending with which buyer, when were they submitted, and what feedback has come back?

  • Development summary for a concept

    Prepare a development summary for this concept from the trial records: what was tried, what worked, and what remains open.

  • Trials that failed and why

    Which trials this year did not proceed, and what reason was recorded for each?

  • Fabric constructions developed

    Which new fabric constructions have we developed this season, and which have moved to bulk?

  • Certification status for a fibre

    What certification is on record for this fibre or finish, what does it cover, and when does it expire?

  • Trials awaiting a decision

    Which completed trials are still awaiting a go or no-go decision, and how long have they been waiting?

R&D: questions before a demo

Our older trial logs and lab-dip cards are on paper. Can it still help?

Only once they are in a file with text. A scanned PDF or a photo of a card has no text to read, so the agent cannot use it. Type the key fields into a Google Sheet or an Excel trial register: trial number, date, buyer, fibre and blend, construction, result, and where the swatch is stored. Keep it on one sheet, one row per trial. Start with the concepts buyers ask for most. Briefs in Word, logs in Excel and test reports as text PDFs can be uploaded as they are, up to 10 MB each.

Can it find the TC that lists a trial’s yarn lot?

Yes, when the TC is a file with text. Upload the transaction certificates to R&D with the trial logs, and ask about a trial. The agent takes the lot from the trial log and quotes the TC on file that lists it: the certificate number, the product, the shipment and the dates. If the log never recorded a lot, or no TC on file lists it, it says so. A scanned certificate has no text to read, so ask the supplier for the PDF. It quotes what the certificate says and does not check it with the certification body; certification scope and chain of custody belong to Compliance & Sustainability.

Will Merchandising or other departments see our buyer-only trials and costs?

Not if the files are uploaded to R&D. Mavick builds that boundary into how it stores and finds data, and tests it with attacks before every release. Buyer-only trials uploaded to R&D stay in R&D; Merchandising’s agent cannot open them, your trial logs or your trial costs. A file shared across the factory is the exception, since other departments can read it, so keep buyer-only work in R&D. A Google Sheet connected to R&D is read by the R&D agent; who can open it in Google stays as you set it there. Mavick does not control the sample room or the swatches in it.

Can the agent suggest a blend, recipe or finish for a new development?

No. Trial recipes, blends and process parameters are values it never makes up, and it does not touch or judge fabric. It shows what your past trials used and recorded, such as neps on a recycled-cotton blend or a hand-feel note against the buyer’s benchmark, each with its source. It can quote the playbook, which says to get the buyer’s hand-feel benchmark before iterating on shade or GSM. The setting for the next trial stays with your developer.

Will it chase buyers for sample feedback on its own?

No. When someone asks, it reads the sample register and lists the developments with no decision, oldest first. If the R&D mailbox in Gmail or Outlook is connected to the department, it also reads the buyer’s emails. It drafts the follow-up, and an admin, usually the R&D head, approves or declines it; only an approved email is sent. It does not watch the inbox or send reminders between questions, and it knows a buyer’s decision only once it is in an email or the register.

Can it compare this trial’s test results with the brief’s targets?

Yes, when the brief and the results are files with text. Upload the brief with its success criteria, test reports from your own lab or a third-party lab as text PDFs, and lab-dip readings, such as ΔE against the standard, exported to Excel or CSV. The agent sets each target beside the result as recorded and cites both. A target with no recorded result is marked as missing, never filled in. It does not connect to the spectrophotometer or any tester, and whether a near miss is good enough stays with your developer and the buyer.