The Daily Letter

Autonomous operations

A supervised digital newsroom team

Sixteen specialised bots handle the repetitive work of research, drafting, publishing, growth and reporting. Every commercial, legal or subscriber-affecting action stops at a human.

Automate repetition, not accountability.

Demonstration mode · simulated data · no bots are connected yet

7

Bots running

4

Awaiting approval

1

Compliance blocks

1

Failed jobs

Autonomy level

Choose how much runs without you

Runs automatically

  • Research approved sources
  • Draft and schedule editions
  • Internal test emails
  • Publish low-risk editions
  • Routine subscriber emails
  • Lead-magnet delivery
  • Referral workflows
  • Sponsor reports
  • Routine support answers
  • Repurpose published content

Always needs your approval

  • New sponsor campaigns
  • Dedicated sponsor emails
  • Financial promotions
  • Legal, political, medical or investment claims
  • Major pricing changes
  • Refunds
  • Deleting user data
  • Account suspension
  • Subscriber-list exports
  • New external integrations

The bot team

Permission map

No bot has unrestricted access. Each one carries its own service account, allowed actions, restricted actions, confidence threshold and escalation rule.

Production

Editor-in-Chief Bot

running

Maintains editorial direction, weekly plans and section balance for each publication.

Allowed
Create editorial plans · Assign bot tasks · Rank stories · Recommend placements
Restricted
Publishing without the autonomy mode allowing it · Pricing changes · Sponsor acceptance
Escalation
Any edition containing a high-risk claim goes to the approval inbox.

Confidence threshold 88%

Production

Research Bot

running

Collects timely material from approved public sources with full provenance.

Allowed
Approved RSS · Permitted APIs · Public pages honouring robots.txt · Internal records
Restricted
Paywall bypass · Authenticated sources · Private profiles · Fabricated sources
Escalation
Unclear or low-quality sources are flagged for human review.

Confidence threshold 82%

Production

Story Scoring Bot

running

Scores every candidate 0–100 with a visible reason for each factor.

Allowed
Score · Rank · Archive below threshold
Restricted
Treating a score as proof of accuracy · Auto-publishing high-risk items
Escalation
Any risk flag forces mandatory human review regardless of score.

Confidence threshold 91%

Production

Duplicate & Fact-Check Bot

running

Detects repetition, recycled releases and unsupported claims.

Allowed
Compare with previous editions · Flag claims · Set verification status
Restricted
Publishing unverified high-impact claims · Removing source attribution
Escalation
Single-source or unverified claims block autonomous publication.

Confidence threshold 86%

Production

Newsletter Writer Bot

waiting

Drafts complete editions in the publication's brand voice and structure.

Allowed
Draft sections · Summarise sourced material · Apply style guide
Restricted
Inventing quotes, statistics, companies or results · Raw subscriber exports
Escalation
Drafts exceeding the risk threshold are routed for approval.

Confidence threshold 84%

Production

Headline & Subject-Line Bot

waiting

Produces at least five scored subject-line options per edition.

Allowed
Generate options · Score clarity, spam risk, mobile length
Restricted
Misleading clickbait · False urgency · Deceptive Re:/Fwd: prefixes
Escalation
Automatic selection only when the Admin enables it.

Confidence threshold 79%

Production

Sponsor Placement Bot

waiting

Places approved sponsor assets with disclosure and exclusivity checks.

Allowed
Approved assets only · Tracking links · Required labels
Restricted
Unapproved creative · Competing sponsors in one edition · Disguising paid content
Escalation
New sponsor campaigns always require Admin acceptance.

Confidence threshold 93%

Publishing

Compliance Bot

blocked

Green / amber / red review of every edition before scheduling.

Allowed
Review · Flag · Block red-rated editions
Restricted
Providing legal advice · Clearing its own amber findings
Escalation
Amber proceeds only with human approval; red blocks publication.

Confidence threshold 90%

Publishing

Quality-Assurance Bot

running

Final inspection for links, placeholders, rendering and footer requirements.

Allowed
Link checks · Render checks · Block sends
Restricted
Overriding a compliance block
Escalation
Missing unsubscribe, placeholder or failed render blocks the send.

Confidence threshold 95%

Publishing

Scheduling Bot

running

Builds the publishing calendar and prevents fatigue and conflicts.

Allowed
Queue jobs · Reschedule failures · Apply frequency caps
Restricted
Changing privacy terms · Sending without approval in Manual/Assisted
Escalation
Calendar conflicts and missed deadlines alert the Admin.

Confidence threshold 92%

Publishing

Autonomous Send Engine

paused

Idempotent test then production sending via the configured provider.

Allowed
Test sends · Consent and suppression checks · Provider dispatch
Restricted
Duplicate sends · Reporting success without provider confirmation
Escalation
Every active send has a kill switch.

Confidence threshold —

Growth

Content Repurposing Bot

waiting

Turns each published edition into social, blog and lead-magnet assets.

Allowed
Draft channel variants · Attach tracked links
Restricted
Publishing externally without a configured integration
Escalation
External posting requires the autonomy mode plus an active integration.

Confidence threshold 80%

Growth

Subscriber Lifecycle Bot

paused

Welcome, preference, re-engagement and unsubscribe journeys.

Allowed
Routine journey emails within frequency caps
Restricted
Re-subscribing an unsubscribed person · Ignoring suppression
Escalation
Any data request or deletion goes to the approval inbox.

Confidence threshold 87%

Revenue

Sponsor Prospecting Bot

paused

Builds scored sponsor prospect lists from public business sources.

Allowed
Public company data · Source URL and retrieval date
Restricted
Guessed emails · Private profiles · Bulk automated sending
Escalation
First outreach to a new segment always requires approval.

Confidence threshold 74%

Operations

Customer-Support Bot

waiting

Routine answers and ticket creation for subscribers, sponsors and managers.

Allowed
FAQ answers · Create tickets · Collect diagnostics
Restricted
Refunds · Deleting accounts · Revealing subscriber or other-tenant data
Escalation
Fraud, breach, refund, dispute or legal threat escalates immediately.

Confidence threshold 76%

Operations

Website Operations Bot

running

Monitors links, forms, webhooks, queues, providers and model health.

Allowed
Retry safe jobs · Disable a failing automation · Raise incidents
Restricted
Rotating financial credentials · Deleting data · Changing permissions
Escalation
Complaint spikes halt sending and notify the Admin.

Confidence threshold 89%

Human approval inbox

Nothing high-risk moves without you

AP-118 · Newsletter edition

SME Growth & Funding Brief

Lead story contains a financial promotion claim requiring human sign-off.

Editor-in-Chief Bot · confidence 90% · high risk

Evidence: 3 primary sources · 1 single-source claim flagged

AP-117 · Sponsor campaign

UK Property & Capital Brief

New bridging lender campaign; category exclusivity conflict with an existing partner.

Sponsor Placement Bot · confidence 93% · high risk

Evidence: Contract dates overlap by 11 days

AP-116 · New outreach segment

Construction Opportunities Brief

First outbound campaign to a newly created 42-company segment.

Sponsor Prospecting Bot · confidence 74% · medium risk

Evidence: All records sourced from public company pages with retrieval dates

AP-115 · Data request

Board & Leadership Brief

Subscriber has requested a copy of their personal data.

Customer-Support Bot · confidence 99% · high risk

Evidence: Identity confirmed via double opt-in address

Task queue

Every automated action is logged

TaskBotActionRiskConfidenceModelStatus
TQ-2418Research BotCollect approved sources for edition 12UK Property & Capital Brief · Monday 06:00 schedule rule · 06:00low82%local: qwen3:8brunning
TQ-2419Story Scoring BotScore 46 candidate storiesUK Property & Capital Brief · Depends on TQ-2418 · —low—local: qwen3:4bqueued
TQ-2411Newsletter Writer BotDraft edition 08Board & Leadership Brief · Editorial plan approved · Tue 09:12medium84%local: mistral-smallwaiting
TQ-2405Compliance BotReview edition 05 — financial promotion riskSME Growth & Funding Brief · Pre-schedule gate · Tue 11:40high90%local: qwen3:14bescalated
TQ-2398Quality-Assurance BotLink and render inspectionConstruction Opportunities Brief · Draft finalised · Tue 08:02low95%rules engine (no AI)failed
TQ-2390Reporting BotSponsor campaign report — edition 03Social Housing Intelligence Brief · 48h post-send · Mon 07:30low97%rules engine (no AI)completed

Default weekly workflow

One edition, five supervised days

  1. Monday

    • Research Bot collects approved stories
    • Story Bot scores candidates
    • Duplicate Bot removes repetition
    • Editor Bot writes the plan

  2. Tuesday

    • Writer Bot drafts the edition
    • Sponsor Bot inserts approved campaigns
    • Headline Bot creates subject options
    • Source Bot validates citations

  3. Wednesday

    • Compliance Bot reviews
    • Quality Bot checks links and rendering
    • Editor Bot finalises
    • Test email generated

  4. Thursday

    • Approval deadline
    • Scheduling Bot queues the edition
    • Send Engine publishes at the configured time

  5. Friday

    • Reporting Bot issues the first report
    • Repurposing Bot creates social and blog assets
    • Analytics Bot recommends changes
    • Sponsor Bot sends campaign updates

Default thresholds

Editable limits, not guarantees

Minimum stories required
5
Minimum average story score
70
Minimum source-quality score
65
Maximum unresolved claims
0
Maximum broken links
0
Maximum send retries
2
Maximum workflow retries
3
Subscriber emails per 7 days
3
Sponsor outreach per working day
20
Follow-ups per prospect
2
Minimum autonomous publishing confidence
90%
Amber / red compliance issues allowed
0 / 0

These values are starting points for a UK B2B newsletter portfolio, not industry guarantees. Local model serving (Ollama), workflow orchestration (n8n), pgvector knowledge bases and email providers all require configuration before any bot on this page performs real work.