AI summarises a Discord export. Accord turns every message into data you can trust.

Pasting a Discord export into ChatGPT gets you a summary. Accord gets you data: every message read, classified against knowledge of your game, and linked back to the conversation it came from.

ChatGPT, Claude, Gemini and Perplexity logos

Black Box Summary

Chatbot

Overall, reactions to Patch 4.2 seem mixed. Some players feel certain changes went too far, while others are enjoying the new content. A few people mentioned bugs. It might be worth keeping an eye on sentiment over the next few weeks!

Read ~4% of the messages

No numbers

No sources

Accord

Structured, Traceable Data

Accord

SUMMARY

48,212 of 48,212 messages read

Players are angry about the Undertow Nerf↗, mostly Vex players↗, and 37 left the server↗ after it. Ranked Rewards↗ is landing well.

↗ Every topic opens its own dashboard, with the messages behind it

PATCH 4.2 · SENTIMENT

14 days since release

−18%

31% positive

20% neutral

49% negative

WHAT’S DRIVING IT

Undertow Nerf

−64%

1,284 msgs

Explore →

Ranked Rewards

+41%

508 msgs

Explore →

Map Rotation

+22%

312 msgs

Explore →

Ranked queue crash

−38%

221 msgs

Explore →

Community Manager

CEO

Senior Dev

Vex Arena · Discord

48,212 messages · 3 channels · last 14 days

What did people think about Patch 4.2? What are they angry about?

DiscordSteamYouTubeReddit

4 sources connected

SAME PROMPT · DIFFERENT RESULTS

So what’s actually different?

ChatGPT, Claude, Gemini and Perplexity

Most studios we speak to have already tried it. Export the Discord, paste it into ChatGPT, Claude or Gemini, and ask what players think. The better setups wire up a bot to do it on a schedule, with a prompt someone has tuned. It’s a real step up from scrolling, and what comes back sounds right: “players want better balance”, “some users report bugs”. But it’s still a summary. The model only reads what fits in its context window, it doesn’t know your game’s characters, items or slang, the topics shift every run so nothing compares one update to the next, and it can’t point to the messages behind a claim. The bot is also one more thing your team has to build, host and keep fixing.

Accord

Accord works the other way round. Every message goes through the same process before anyone asks a question.

Every message is read. Pulled in from your Discord on a regular sync, with each reply kept in its conversation, not sampled to fit a prompt.

Matched to your game. Classified against knowledge of your characters, items, modes, systems and the nicknames players give them.

Tagged by who’s talking. Sorted by cohort: new players, veterans, creators and the rest of your community.

Counted and traced. Turned into numbers you can compare update to update, each linked back to the messages behind it.

How every message is classified

Accord reads every message the way your best community manager would. It follows replies, recognises your game’s characters, items and updates, scores sentiment and connects it all to who’s talking, then turns it into numbers you can track.

# balance-discussionTOPICUndertow nerf backlashLive example · click any highlighted word
AccordBOT
Reading every channel in Vex Arena
182,430
messages read
mapleflowerToday at 6:12 PM
They gutted in fr
Today at 6:13 PM
Literally
1 of 6 · Thread
Finds the reply
Context recovered · 7d
Replies4,210
Quoted messages1,530
Reaction-only880
Inherited topicsPatch 4.2Undertow NerfUndertow

“Literally unplayable” has no subject on its own. Accord follows the reply and inherits what it is about.

In the weekly report
Counted as Undertow Nerf feedback, not lost as an unlabelled one-liner.
What Accord records from these two messages

Book a Demo

how Accord knows your game

Accuracy comes from what happens before anyone asks a question. Before your first login, Accord has built a model of your game and run your community’s history through it. Every new message goes through the same process.

Knowledge of your game

Accord knows the topics of conversation and the context behind conversations to make accurate groupings

KNOWLEDGE SOURCES

GAME KNOWLEDGE

DISCUSSION TOPICS

Steam

Steam

News

Update history · Events

Fandomwiki.gg

Wikis · Characters · Items

Discord

Discord

#announcements

Patch notes · Dev updates

Knowledge base

Docs · Notes · Roadmaps

Accord

Knowledge of your game

reading sources…

✓ knowledge up to date

Balance

Bugs

Updates

Undertow Nerf

Mechanic

186 msgs

Frost Staff

Item

94 msgs

Login queue

Known issue

121 msgs

Crafting dupe

System

64 msgs

Patch 4.2

Update

142 msgs

KNOWLEDGE SOURCES

Steam

Steam

Update history · Events

Fandomwiki.gg

Wikis · Characters · Items

Discord

Discord

Patch notes · Dev updates

Knowledge base

Docs · Notes · Roadmaps

Accord

Knowledge of your game

reading sources…

✓ knowledge up to date

DISCUSSION TOPICS

Balance

Undertow Nerf

Mechanic

186 msgs

Frost Staff

Item

94 msgs

Bugs

Login queue

Known issue

121 msgs

Crafting dupe

System

64 msgs

Updates

Patch 4.2

Update

142 msgs

Our engine to process your messages

The same steps run on your community’s history and on every new message as it arrives.

Pre-processing. Replies are joined to what they reply to, bots and spam are removed, and languages are detected, so a three-word Discord message keeps its context.

Game knowledge. Accord reads your patch notes, wiki, store pages and update history, so it knows what shipped and when.

Your game’s vocabulary. From those sources and the community itself, Accord learns your characters, items, modes, maps and systems, and the slang and nicknames players use for each.

Classification. Every message is classified by topic, sentiment, bug and request against a taxonomy built for your game rather than a generic one.

Outputs. Counts, trends, cohorts and alerts you can filter and compare across updates. Every number links back to the messages behind it.

30-minute demo · your community history analysed from day one

A Toolkit to listen to your players

Accord isn’t a chatbot with your Discord attached. It’s an engine that runs on every message, all the time, so community, live-ops, design and studio leadership can make confident decisions about where the game goes next.

Bring every source together

Connect

At the centre of your stack

Accord connects to where your players talk and where your team works: Discord, Steam, Fandom, Slack, Linear, Jira, and AI assistants like Claude over MCP.

See it on your community

Turn messages into meaning

Understand

Every message, labelled

Each message is scanned for the things it mentions, then given a topic, intent and sentiment, so the picture doesn’t depend on what someone happened to read.

See it on your community

Watch your community change

Track

See how your community is doing

Server growth, engagement, sentiment, retention and activation, every period against the last, with who returned, who went quiet and who spoke for the first time.

See it on your community

Move before it spreads

Act

Spot issues as they emerge

The right people are notified about trends, bugs and requests on the topics they own, in Slack, Discord or email.

See it on your community

Every message

Read and classified continuously, not sampled to fit a prompt

Unique to your game

Accord knows your characters, items and modes, and the slang players use for them.

Data you can trust

Linked to the messages behind it, so you can check the work and quantify drivers in your community.

“That’s a full-time position on its own. To catch up with everything being said in the community, that’s a task with a lot of manual input, a lot of reading, sometimes repeating. When you are focused on one conversation, you sometimes miss the bigger picture. Accord brings the capability of seeing things with a wider perspective.”

GOALS

Adrián Álvarez Martínez, Community & Influencer Manager, GOALS

Easy to check so you don’t have to take our word for it

The first question on almost every demo is “how accurate is it?” You shouldn’t have to take our word for it. Every summary, chart and alert in Accord links to the source messages behind it, so anyone on the team can open the evidence and decide for themselves.

“Everybody [in our team] has access to our reports. Everybody can check them, everybody can have an opinion. There’s no gatekeeping of information.”

GOALS

Adrián Álvarez Martínez, Community & Influencer Manager, GOALS

When something is wrong, you fix it where you see it. Reassign a message to the right topic and Accord learns from the correction straight away. Slang, sarcasm and in-jokes are why Accord learns your game first: one community we work with reduces almost everything to three-letter shorthand that means nothing to an outsider, and Accord still understands it.

At the volume a live game produces, a handful of misread messages doesn’t change a trend. Missing most of the conversation does.

Feature

AI Summaries

Accord vs ChatGPT, Claude, Gemini, Grok

GETTING THE DATA

Storage

Discord connection

Data in

Always synced, structured data

Always synced, structured data

A bot your team has to build, host and keep fixing

A bot your team has to build, host and keep fixing

Storage

Storage

Millions of messages saved, every member’s history kept

Millions of messages saved, every member’s history kept

Gone when the chat ends

Gone when the chat ends

Volume

Volume

Your whole community, from your chosen sync date

Your whole community, from your chosen sync date

Whatever fits in the context window

Whatever fits in the context window

Bots and staff

Bots and staff

Identified, grouped and excluded from every metric, so you get a clear player signal

Identified, grouped and excluded from every metric, so you get a clear player signal

Mixed in with players

Mixed in with players

UNDERSTANDING IT

Storage

Message labels

Message labels

Topic, intent and sentiment on every message

Topic, intent and sentiment on every message

✗

✗

Knows your game

Knows your game

Taxonomy built from your knowledge articles: items, characters, creators, updates, announcements, events and known issues, used on every message

Taxonomy built from your knowledge articles: items, characters, creators, updates, announcements, events and known issues, used on every message

Doesn’t know your mechanics, items, characters or story. With no source of truth, it fills the gaps with guesses

✗

Topics

Topics

Trackable topics driven by your game’s taxonomy and real conversations

Trackable topics driven by your game’s taxonomy and real conversations

Change every run

Change every run

Summaries

Summaries

Built from classified conversations across your whole history, with every point linked to its messages

✓

Of whatever you paste in, read as one long block

✓

TRUSTING IT

Storage

Counts

Counts

Exact

Exact

Guessed

Guessed

Sources

Sources

Every claim is checked against the real messages before it ships, then linked straight to the message in Discord

Every claim is checked against the real messages before it ships, then linked straight to the message in Discord

Can’t link back to the source message, and often hallucinates quotes or figures

Often hallucinates quotes or figures, and you can’t tell which

Auditable

Auditable

Every number traces back to the messages behind it

Every number traces back to the messages behind it

A black box

A black box

ACTING ON IT

Storage

Cohorts

Cohorts

Segment by Discord roles (platform, region, playstyle or anything from onboarding), tenure and activity, plus topic, intent and sentiment

Rule-based audiences: any mix of topic, intent, sentiment, topic focus, message count, channel, role, join date and leave date

✗

✗

Activity

Activity

Server growth, engagement, sentiment, retention and activation, every period against the last, with who returned, who went quiet and who spoke for the first time

Server growth, engagement, sentiment, retention and activation, every period against the last, with who returned, who went quiet and who spoke for the first time

✗

✗

Bug inbox

Bug inbox

Bugs found, grouped and prioritised as they come in, then traced back through millions of messages to every report of the same issue

Bugs found, grouped and prioritised from player messages as they come in

✗

✗

Alerts and digests

Alerts and digests

The right people or teams notified about trends, bugs and requests on the topics they own, in Slack, Discord or email

Bugs and insights sent to Slack, Discord or email as they happen

✗

✗

Reports

Reports

Built in: layer any filters (topic, intent, channel, cohort, keyword) on one live chart, break it down, trend it, always up to date

Built in: layer any filters (topic, intent, channel, cohort, keyword) on one live chart, break it down, trend it, always up to date

Without consistent tagging there’s nothing to count or compare, so it’s spreadsheets by hand

Built by hand in spreadsheets or separate tooling

Exports

Exports

PDF or Markdown in one click, or straight into Linear as issues

PDF or Markdown, one click

Copy and paste into your tracker

Copy and paste

WORKING WITH ACCORD

Storage

Tuned to you

Tuned to you

Classifications you correct stick, and we tune the taxonomy to your preferences

Classifications you correct stick, and we tune the taxonomy to your preferences

Same answer for every game

Same answer for every game

Service

Service

White-glove: our team sets it up, watches quality and keeps it right

White-glove: our team sets it up, watches quality and keeps it right

You’re on your own

You’re on your own

Cost

Cost

Often cheaper

Often cheaper

Tokens, staff time, external reporting software and downtime

Tokens, staff time, external reporting software and downtime

Frequently Asked Questions

How accurate is it?
Why not just do this ourselves with ChatGPT or Claude?
Our players talk in slang, sarcasm and in-jokes. Can it keep up?

Your community already knows your next product.

Accord reads every message in your community, filters the noise and tells you what your players want before your competitors work it out.

Accord reads every message in your community, filters the noise and tells you what your players want before your competitors work it out.

By submitting, you agree to our privacy policy.

No commitment. We'll reply within one business day.

Hear from our customers

One Community Manager vs One Million Players on Launch Day

One Community Manager vs One Million Players on Launch Day

“When you are focused on one conversation, you sometimes miss the bigger picture. Accord brings the capability of seeing things with a wider perspective.”

Adrián Álvarez Martínez

Community & Influencer Manager, GOALS

Read

How Essential Mod turned community insight into 3 record-breaking collections

How Essential Mod turned community insight into 3 record-breaking collections

“Had I just looked at my Discord server and gone to the team, then maybe we would have made an ill-advised change.”

Greg Castle

Growth Community Manager, Essential Mod

Read