This Week at UnaGo
It's been a big week. We've partnered with Doubleword AI to bring cost-effective AI inference to UK and European businesses, shipped support for Midjourney 8.1, shipped native integrations for Google's marketing stack, Ahrefs, Slack, Monday.com, and HubSpot - connecting your tools directly to specialist AI agents.
Here's everything that shipped.
Partnership: Doubleword AI and Low-Cost AI for UK and EU Teams
We're proud to announce our partnership with Doubleword AI, one of the UK's first dedicated inference labs, to deliver low-cost AI capabilities to UK and European businesses.
Doubleword is backed by the UK government's Sovereign AI fund - a strategic investment programme designed to build domestic AI capability and reduce reliance on overseas infrastructure. By partnering with a government-backed inference lab, UnaGo users get access to infrastructure that's part of a national strategy to keep AI innovation, talent, and compute within the UK and Europe.
Why This Matters
The UK and EU have committed significant investment to support domestic AI startups and build sovereign AI infrastructure. Doubleword, led by CEO Meryem Arik, is at the forefront of this movement - focused on making AI inference dramatically more cost-effective for businesses across the region.
The challenge is clear: running large language models at scale is expensive. Inference costs - the computational expense of generating AI responses - remain a major barrier for businesses scaling AI operations. Doubleword's approach tackles this head-on with optimised inference pipelines that reduce token costs without sacrificing quality.
What This Means for UnaGo Users
Through this partnership, UnaGo users will benefit from:
- Lower operational costs for high-volume AI workflows
- Faster inference times on complex multi-agent tasks
- UK and EU data residency for regulated industries and data-sensitive businesses
- Sustainable scaling - AI that doesn't break the budget as usage grows
This aligns with the UK and EU's broader AI strategy: building domestic capability, retaining talent from institutions like Oxford, UCL, and ETH Zurich, and creating a regulatory environment that enables innovation rather than stifling it.
For teams working with large datasets or running high-frequency automated workflows, this partnership unlocks AI adoption at a price point that finally makes sense.
Learn more about UnaGo's platform and how we're making AI operations accessible.
Product Update: Midjourney 8.1 Model Support
We've added full support for Midjourney 8.1, the latest iteration of Midjourney's image generation models.
What's New in 8.1
Midjourney 8.1 brings improvements to:
- Photorealism - More natural lighting, textures, and human anatomy
- Style consistency - Better coherence across multi-image campaigns
- Prompt adherence - More accurate interpretation of complex creative briefs
- Speed - Faster generation times for production workflows
How UnaGo Uses Midjourney 8.1
Our specialist agents can now leverage Midjourney 8.1 for:
- Marketing campaigns - Generate on-brand social media graphics, ad creatives, and campaign assets
- Content production - Create featured images, infographics, and visual storytelling elements
- Product visualisation - Produce concept art, mockups, and design explorations
- Brand consistency - Maintain visual identity across hundreds of generated assets
When you brief a Production Agent with "Create 5 LinkedIn graphics for our new product launch using our brand colours", it now taps into Midjourney 8.1's improved capabilities to deliver higher-quality results faster.
See how AI agents handle creative production.
Agentic Tuning: Manual and Automated Improvements
We've shipped significant improvements to how agents learn and adapt to your specific workflows.
Manual Tuning Enhancements
You can now:
- Fine-tune agent behaviour with more granular control over response patterns
- Adjust reasoning depth - choose between fast, surface-level responses and deep, multi-step analysis
- Customise tool selection - prioritise which integrations agents use for specific task types
- Set quality thresholds - define minimum standards before agents deliver work products
Automated Tuning Improvements
Our orchestration layer now:
- Learns from feedback - agents improve based on your approvals, edits, and rejections
- Optimises tool chains - automatically identifies the most efficient sequence of tools for recurring tasks
- Balances cost and quality - dynamically adjusts inference parameters based on task complexity
- Detects drift - flags when agent performance degrades and suggests recalibration
The result? Agents that get smarter over time without requiring constant manual intervention. You set the goals, and the system figures out the most efficient way to achieve them.
Explore agent orchestration capabilities.
New Native Integrations: Google Stack, Slack, Ahrefs, and More
We've shipped a wave of native integrations that connect your marketing, sales, and operations tools directly to UnaGo's specialist agents - no middleware, no manual exports, no context switching.
Google Marketing Stack
Full native support for Google's three core marketing tools:
Google Search Console - Monitor indexing status, pull impressions, clicks, CTR, and position data, submit sitemaps, and inspect URLs to diagnose ranking issues.
Google Analytics 4 - Pull traffic reports, analyse user behaviour, segment audiences, and track campaign performance across channels.
Google Ads - Monitor campaign performance, analyse keyword effectiveness, generate optimisation recommendations, and pull quality scores.
The real power comes from combining them in one conversation:
"Pull our top 10 landing pages from GA4 by conversion rate, check their Search Console performance for the keywords we're bidding on in Google Ads, and suggest which organic pages we should promote with paid campaigns."
No spreadsheets. No manual exports. All in one conversation.
Slack Bot: Two-Way AI Communication
Our native Slack bot brings UnaGo's specialist agents directly into your team's workspace. Mention @UnaGo in any channel to brief an agent, and agents post completed work directly to specified channels.
Marketing teams: "@UnaGo, monitor our brand mentions on Twitter today and post a summary to #marketing at 5pm"
Sales teams: "@UnaGo, when a new lead fills out our demo request form, research their company and post a briefing to #sales-intel"
Operations teams: "@UnaGo, every Monday at 9am, pull our support ticket metrics and post a weekly report to #ops"
No context switching. No friction. Just execution.
SEO and Advertising Intelligence
Ahrefs - Enterprise SEO intelligence: batch URL analysis, backlink profiles, organic traffic data, and brand radar tracking (AI responses mentioning your brand or competitors).
Facebook Ads - Create ads and creatives, manage campaigns, and research competitor ad strategies via Ad Library integration.
Foreplay - Competitive ad research: find ads using the same creative assets across brands and campaigns.
Project Management and CRM
Monday.com - Full board, item, and group management with sprint planning and workflow automation.
Asana - Task creation, project management, and team collaboration across workspaces.
HubSpot - Contact, company, and deal management with pipeline automation and CRM property updates.
Salesforce - Data warehouse integration for cross-referencing CRM data with revenue, support, and product analytics.
Data Warehouse Connectors
Connect your operational data sources for unified analytics:
- Stripe - Revenue and payment data (with real-time webhook sync)
- Shopify - E-commerce transactions and customer data
- Snowflake and BigQuery - Cloud data warehouse queries
- Postgres and MySQL - Direct database connections
- Zendesk - Support ticket analytics
Each integration is fully discoverable by specialist agents - meaning you can say "Pull our Q2 sales data from Stripe, analyse trends, and create a report" without configuring anything manually.
See how UnaGo handles marketing analytics.
Coming Next: Async AI Workflows with Doubleword
We're building on our Doubleword partnership to introduce cost-efficient asynchronous AI workflows - designed for large-scale data processing that doesn't need instant results.
The Problem
Some AI tasks are massive:
- Analysing 10,000 customer support transcripts for sentiment trends
- Processing a year's worth of sales calls to identify winning patterns
- Generating personalised outreach for 5,000 leads
- Translating and localising 500 pages of documentation
Running these in real-time is expensive and often unnecessary. You don't need the results in 30 seconds - you need them by tomorrow morning, at a price that makes sense.
The Solution: Async Workflows
With Doubleword's optimised inference infrastructure, we're introducing async job queues:
- Submit large jobs - Brief an agent on a massive task
- Queue and process - Work runs in the background using cost-efficient inference
- Deliver when ready - Results posted to Slack, email, or your preferred channel
- Pay less - Async processing costs 60-80% less than real-time inference
Use Cases
Content teams:
- "Analyse our last 200 blog posts and identify which topics drove the most organic traffic. Deliver a report by Friday."
Sales teams:
- "Research all 3,000 leads in our CRM, score them by fit, and create personalised outreach drafts for the top 500. Ready by Monday."
Support teams:
- "Read every support ticket from Q1, categorise issues, and identify the top 10 product improvements that would reduce ticket volume. Report due next week."
Research teams:
- "Monitor 50 competitor blogs for the next month, summarise new posts, and flag anything we should respond to. Weekly digest every Friday."
Cost Efficiency at Scale
The economics are simple:
- Real-time inference: £0.05-0.10 per complex task
- Async inference: £0.01-0.03 per complex task
For a 5,000-lead research job:
- Real-time: £250-500
- Async: £50-150
Same quality. Same agents. Just smarter scheduling.
This is how AI becomes sustainable for large-scale operations - not by cutting corners on capability, but by matching the right compute strategy to the right job.
Join the waitlist for async workflows.
What's Next
We're just getting started. Here's what's on the roadmap:
- Expanded async capabilities - More workflow types optimised for cost efficiency
- Advanced agent specialisation - New agent types for legal, finance, and HR workflows
- Deeper Google integration - Google Workspace automation (Docs, Sheets, Slides)
- Enterprise features - SSO, audit logs, custom agent training, dedicated infrastructure
The execution gap is closing. Teams that adopt AI-native operations now will outpace those still stuck in manual workflows.
See what UnaGo can do for your team.
Stay Connected
Questions? Feedback? Reply to this post or reach out to our team. We read everything.