AI News Digest, April 28, 2026: An AI Voice Agent For Plumbers Just Hit $1B

Three things landed in the AI agents enterprise stack this week, and they do not fit the usual narrative. A startup answering missed calls for plumbers just hit unicorn status. OpenAI rebuilt custom GPTs into shared agents that live inside Slack and Salesforce. And in Brussels today, EU negotiators are voting on whether to push the AI Act’s hardest deadlines back by more than a year. Meanwhile, fresh PwC data says 76% of enterprises using AI in HR have not started compliance prep. The agentic layer is moving faster than the rulebook. Here is what changed and what to do about it.

An AI Voice Agent For Plumbers Just Hit $1B. Welcome To The Vertical Layer.

On April 27, Avoca announced it raised more than $125 million across Seed, Series A, and Series B at a $1 billion valuation. Series B was led by Meritech and General Catalyst. Series A was led by Kleiner Perkins, with backing from Amplify Partners and Y Combinator. (Source: Fortune, PR Newswire.)

The company sells AI voice and workflow agents to operators in HVAC, plumbing, automotive, moving, and other field-services categories. The agent answers every inbound call within seconds. Then it books jobs into the customer’s CRM. After that, it follows up on outstanding estimates. Finally, it dynamically reroutes lead flow based on technician capacity. Avoca says it is on track to book $1 billion in jobs this year.

Why this matters for founders and HR leaders

For two years, the conversation about AI agents was dominated by horizontal players. ChatGPT, Claude, copilots that did everything for everyone. However, Avoca shows the next wave: vertical agents trained on the actual workflow of one industry, sold to small operators who never considered themselves AI buyers. A plumber in Phoenix is now an enterprise AI customer.

For founders, that opens two playbooks. First, vertical AI agents in services categories are getting funded at unicorn multiples even when the underlying business is unglamorous. Second, the talent fight to build them is real. Avoca’s hiring page lists roles in San Francisco at frontier-lab compensation. So if you are recruiting for an applied AI startup, your competitor is no longer just Anthropic and OpenAI. It is also a Y Combinator company selling to roofers.

For HR leaders inside service-economy companies, this is a quieter shift. The field tech you hire next year may show up on day one with an AI dispatcher already booking their calendar. As a result, onboarding has to account for that. So does performance management.

What to do this week

First, if you run a services or field-ops business, ask your CRM vendor what voice-agent integrations they support. Second, if you run HR for one, talk to ops leadership about which AI agents are already touching scheduling and dispatch. You cannot manage performance for tools you do not know are there. Specifically, AI agents in HR workflows are following the same vertical pattern.

EU AI Act Trilogue Votes Today On Whether To Push Hard Deadlines To 2027

EU negotiators meet in Brussels for the final political trilogue on the Digital Omnibus on AI today, April 28. Both Parliament and Council reject the Commission’s conditional delay. Instead, they replace it with hard dates. Specifically, December 2, 2027 covers stand-alone Annex III high-risk systems. Meanwhile, August 2, 2028 covers high-risk AI embedded in regulated products. Formal adoption is expected by July. (Source: Iubenda, IAPP.)

So what? Maybe you hire in the EU. Maybe you run payroll for EU employees. Or you use AI for resume screening or video interviews touching EU candidates. Either way, you have been building toward an August 2026 deadline. After today’s vote, that probably slips by 16 months. However, the substance does not get easier. The conformity assessment, technical documentation, and CE marking obligations all stay. You buy time, not a pass. Therefore, founders should use the runway to actually finish the AI vendor inventory rather than treat the delay as relief.

OpenAI’s Workspace Agents Move The Battle Inside Slack And Salesforce

On April 22, OpenAI launched Workspace Agents in ChatGPT, the explicit successor to custom GPTs for organizations. The agents are powered by Codex, run in the cloud, and plug into Slack, Google Drive, Microsoft 365, Salesforce, Notion, and Atlassian Rovo. Pricing is free until May 6, then moves to credit-based billing on Business and Enterprise plans. (Source: VentureBeat, 9to5Mac.)

For the AI agents enterprise stack inside white-collar teams, this is the inverse of the Avoca move. Instead of going vertical, OpenAI is going horizontal. Specifically, it embeds directly inside the apps your team already pays for. For example, your recruiting team uses Slack, Notion, and Greenhouse. A Workspace Agent can now read a hiring channel, summarize candidate threads, and draft interview prep. Nobody leaves the chat. However, there is a catch. Custom GPTs are being deprecated for organizations. Anyone who built workflows on the old standard has to migrate. Therefore plan that audit before May 6 to avoid surprise charges. Founders evaluating AI tools HR teams trust should treat Workspace Agents as a replacement for ad hoc GPTs, not a new category.

PwC: Three In Four Enterprises Using AI In HR Have Not Started EU Compliance Prep

PwC reports that only 24% of enterprises using AI in HR have begun formal EU AI Act compliance preparation. The rest are deploying AI in recruitment, screening, and workforce analytics without a documented compliance plan. Maximum fines run from €7.5 million or 1.5% of global turnover up to €35 million or 7% of global turnover, depending on the violation. (Source: Intervue, PwC.)

For HR leaders, the gap is the story. The Act classifies almost every AI tool used in hiring as high-risk. That covers ATS ranking engines, interview scorers, and the workflows around them. Even with the likely Brussels delay, the audit work does not go away. Instead, it just shifts. For instance, you run a 200-person startup with a distributed EU team. The cheapest move is a written inventory. List every AI vendor your recruiters touch, the model behind it, and the bias-monitoring evidence each vendor will produce on demand. Most enterprises do not have this list. Moreover, closing the AI skills gap in HR means giving the team the literacy to even ask vendors the right questions.

Quick Hits

• Anthropic’s Project Deal: Claude agents acting for 69 employees closed 186 deals worth over $4,000 in a one-week internal marketplace experiment. Smarter models won every time. (TechCrunch)
• Korea-Japan Twin Hub AI alliance: Business leaders proposed splitting AI training in Japan, where power and cooling are favorable, and inference in Korea, which leads in memory chips. (Seoul Economic Daily)
• DeepSeek cuts API cache-hit prices to one-tenth: Effective immediately across all models, with a 75% V4-Pro discount through May 5. As a result, frontier-lab pricing pressure intensifies. (The Next Web)

Maybe today’s news has you rebuilding the AI agents enterprise stack inside your HR function. If so, Asanify’s HRMS platform already supports API-first integration with the agent layer your team is testing. Worth a look before the next budget cycle.

FAQ

Q: What is the AI agents enterprise stack?
The AI agents enterprise stack is the layer of AI tools that act on behalf of employees inside your business apps. For example, voice agents answer calls. Meanwhile, workspace agents work inside Slack, Salesforce, and Notion. It includes vertical agents like Avoca for trades and horizontal ones like OpenAI Workspace Agents for white-collar teams.

Q: Will the EU AI Act August 2026 enforcement deadline really get pushed?
The April 28, 2026 trilogue is voting on hard delay dates. Specifically, December 2, 2027 covers stand-alone high-risk systems. Meanwhile, August 2, 2028 covers AI embedded in regulated products. Formal adoption is expected by July 2026. However, the substantive obligations stay the same, only the timing shifts.

Q: What should HR teams do about EU AI Act compliance right now?
First, start with a written inventory. List every AI vendor your recruiters touch. Note the model each one uses. Capture what bias-monitoring evidence the vendor will produce on demand. PwC reports only 24% of enterprises using AI in HR have begun this work. Therefore, the audit gap is wide and any progress moves you ahead of peers.

Not to be considered as tax, legal, financial or HR advice. Regulations change over time so please consult a lawyer, accountant  or Labour Law  expert for specific guidance.