Contents
We measured which outsourcing companies AI assistants actually recommend: 1,485 real user answers from ChatGPT and Google AI, plus 480 controlled queries we ran across ChatGPT, Claude, Gemini and Perplexity on August 3, 2026. One answering service is named in 85.1 percent of runs. Most of the industry, including us, is invisible.
When a small business owner asks an AI assistant “best answering service for small business” or “top virtual assistant companies”, the answer is a shortlist of three to six names. There is no page two. Whoever is on that shortlist gets the customer, and nobody publishes who that is.
So we measured it. This is the first edition of a quarterly index covering 74 virtual assistant, answering service and SMB outsourcing providers. The method and its limits come first, then the leaderboards.
Method, and what this index cannot tell you
Every number in this report comes from one of two named sources, and the two are never blended.
- Real user answers. 1,485 answers that ChatGPT (226 answers) and Google AI Overviews gave to real user questions about hiring these services, collected from the DataForSEO LLM Mentions database, United States, English, August 2025 through August 2026. These are questions real people asked, not prompts we wrote.
- A controlled panel. 40 buyer questions, drawn from the real-question corpus by search volume (6 Philippines-focused prompts added by us and disclosed as such), each asked 3 times to the current consumer models of ChatGPT (gpt-5.5), Claude (claude-sonnet-4-6), Gemini (gemini-3.5-flash) and Perplexity (sonar-pro) with web search on: 480 runs, all on August 3, 2026.
The counting rule matters. A company scores only when its name appears in the body of an answer, after every citation link is stripped out. Being cited as a source is not a recommendation, and in this data the two diverge constantly. Each company counts at most once per answer, and “average position” records how early in the shortlist it appears, where 1 means named first.
Limits you should hold this against:
- Rankings are within category only. The three categories have different denominators (715 answering answers against 189 virtual assistant answers in the real-user corpus), so a percentage in one table cannot be compared to a percentage in another.
- We excluded three groups by rule, not by judgment call. Enterprise BPO and IT majors (Teleperformance, Concentrix, TaskUs, Accenture and peers), because they do not serve the small business buyer these questions come from. Freelance marketplaces (Upwork, Fiverr, OnlineJobs.ph), because they are platforms rather than managed providers. Software and AI receptionist products, which we track separately below rather than rank against human services.
- AI answers drift. The panel is a snapshot of one day. That is why this is a quarterly index with a frozen dataset per edition, not a live tracker.
- Three runs per prompt is a floor, not a census. Rates for firms below roughly 5 percent are noisy and we treat them as “rarely named” rather than precisely measured.
- We are in this index. Armasourcing ranks near the bottom of the virtual assistant table, and that number is published unedited below. An independent tracker, ZeroRank, shows the same result for us, which is a useful check that our counting is not flattering anyone.
Answering services: one firm owns the AI shortlist
Smith.ai is named in 85.1 percent of all controlled runs, including 97.6 percent of Gemini runs, the highest single number in this study. Ruby and AnswerConnect form a stable second tier. Everyone else is fighting for the occasional sixth slot.
| # | Company | Panel | ChatGPT | Claude | Gemini | Perplexity | Real-answer share | Avg position |
|---|---|---|---|---|---|---|---|---|
| 1 | Smith.ai | 85.1% | 90.5% | 83.3% | 97.6% | 69.0% | 22.38% | 2.83 |
| 2 | Ruby | 69.6% | 88.1% | 66.7% | 66.7% | 57.1% | 28.95% | 2.43 |
| 3 | AnswerConnect | 67.3% | 85.7% | 69.0% | 61.9% | 52.4% | 26.29% | 2.66 |
| 4 | Abby Connect | 31.0% | 31.0% | 19.0% | 38.1% | 35.7% | 9.93% | 3.62 |
| 5 | MAP Communications | 26.2% | 11.9% | 21.4% | 35.7% | 35.7% | 13.57% | 4.25 |
| 6 | AnswerFirst | 23.8% | 31.0% | 42.9% | 4.8% | 16.7% | 6.57% | 3.5 |
| 7 | PATLive | 23.2% | 26.2% | 11.9% | 28.6% | 26.2% | 9.09% | 4.4 |
| 8 | Moneypenny | 20.8% | 26.2% | 7.1% | 9.5% | 40.5% | 8.39% | 3.23 |
| 9 | ReceptionHQ | 19.6% | 2.4% | 19.0% | 14.3% | 42.9% | 23.08% | 3.61 |
| 10 | Davinci Virtual | 18.5% | 21.4% | 11.9% | 4.8% | 35.7% | 3.92% | 4.68 |
| 11 | VoiceNation | 11.9% | 9.5% | 0.0% | 7.1% | 31.0% | 4.48% | 5.1 |
| 12 | Posh | 10.7% | 7.1% | 11.9% | 9.5% | 14.3% | 13.99% | 6.22 |
| 13 | Specialty Answering Service | 9.5% | 19.0% | 2.4% | 2.4% | 14.3% | 10.77% | 2.88 |
| 14 | AnswerForce | 9.5% | 14.3% | 21.4% | 2.4% | 0.0% | 5.31% | 3.28 |
| 15 | AMBS Call Center | 9.5% | 0.0% | 19.0% | 0.0% | 19.0% | 6.99% | 4.16 |
The real-user corpus tells the same story with one twist: Ruby leads the historical record at 28.95 percent of 715 real answers against Smith.ai’s 22.38 percent. Smith.ai’s dominance is strongest in the newest, controlled data. The shortlist is consolidating around it.
Virtual assistant companies: a two-firm race with a crowded middle
BELAY and Time Etc tie at 55.8 percent of runs, with MyOutDesk at 54.5 percent. BELAY is also named earliest: position 2.09 on average when it appears, the strongest lead-off spot in the study. In the real-user corpus BELAY appears in 33.86 percent of answers, the highest share any firm holds in its own category’s historical record.
| # | Company | Panel | ChatGPT | Claude | Gemini | Perplexity | Real-answer share | Avg position |
|---|---|---|---|---|---|---|---|---|
| 1 | BELAY | 55.8% | 64.1% | 46.2% | 53.8% | 59.0% | 33.86% | 2.09 |
| 2 | Time Etc | 55.8% | 69.2% | 30.8% | 64.1% | 59.0% | 24.87% | 3.3 |
| 3 | MyOutDesk | 54.5% | 43.6% | 51.3% | 56.4% | 66.7% | 19.05% | 3.53 |
| 4 | Boldly | 44.9% | 23.1% | 41.0% | 56.4% | 59.0% | 22.22% | 3.27 |
| 5 | Wing Assistant | 44.9% | 38.5% | 28.2% | 56.4% | 56.4% | 17.99% | 4.06 |
| 6 | Prialto | 43.6% | 51.3% | 43.6% | 28.2% | 51.3% | 24.34% | 4.13 |
| 7 | Wishup | 33.3% | 20.5% | 46.2% | 35.9% | 30.8% | 12.17% | 4.77 |
| 8 | Athena | 28.8% | 2.6% | 28.2% | 51.3% | 33.3% | 4.23% | 4.16 |
| 9 | Fancy Hands | 28.8% | 35.9% | 20.5% | 17.9% | 41.0% | 14.29% | 5.67 |
| 10 | VirtualStaff.ph | 21.2% | 15.4% | 23.1% | 23.1% | 23.1% | 4.23% | 2.74 |
| 11 | WoodBows | 17.3% | 28.2% | 20.5% | 2.6% | 17.9% | 5.82% | 6.22 |
| 12 | TaskBullet | 16.0% | 5.1% | 12.8% | 20.5% | 25.6% | 4.76% | 3.93 |
| 13 | ClearDesk | 16.0% | 20.5% | 15.4% | 5.1% | 23.1% | 1.06% | 6.64 |
| 14 | Virtual Coworker | 14.7% | 17.9% | 17.9% | 12.8% | 10.3% | 2.65% | 3.09 |
| 15 | Virtudesk | 13.5% | 17.9% | 7.7% | 7.7% | 20.5% | 5.29% | 3.14 |
SMB outsourcing: the most open category
No firm clears 40 percent. Helpware leads at 39.7 percent, and when it is named it comes early, position 2.12 on average. Philippines-based staffing firms (Cloudstaff, MicroSourcing, Outsourced.ph, Emapta) all register, but none has consolidated the “offshore staffing” question the way Smith.ai has consolidated answering. This is the category where AI visibility is still up for grabs.
| # | Company | Panel | ChatGPT | Claude | Gemini | Perplexity | Real-answer share | Avg position |
|---|---|---|---|---|---|---|---|---|
| 1 | Helpware | 39.7% | 38.5% | 38.5% | 38.5% | 43.6% | 7.06% | 2.12 |
| 2 | SupportYourApp | 30.1% | 20.5% | 43.6% | 20.5% | 35.9% | 6.39% | 2.59 |
| 3 | Hugo | 22.4% | 15.4% | 23.1% | 15.4% | 35.9% | 5.21% | 2.78 |
| 4 | Cloudstaff | 17.9% | 25.6% | 12.8% | 10.3% | 23.1% | 2.69% | 2.84 |
| 5 | Influx | 16.0% | 7.7% | 2.6% | 28.2% | 25.6% | 0.84% | 2.48 |
| 6 | SupportNinja | 16.0% | 38.5% | 15.4% | 2.6% | 7.7% | 2.52% | 2.73 |
| 7 | MicroSourcing | 13.5% | 15.4% | 7.7% | 7.7% | 23.1% | 1.01% | 2.96 |
| 8 | Unity Communications | 12.8% | 20.5% | 2.6% | 12.8% | 15.4% | 3.87% | 3.59 |
| 9 | PartnerHero | 11.5% | 17.9% | 7.7% | 7.7% | 12.8% | 0.67% | 2.94 |
| 10 | Outsourced.ph | 10.9% | 10.3% | 10.3% | 10.3% | 12.8% | 0.5% | 3.11 |
| 11 | Magellan Solutions | 9.0% | 10.3% | 7.7% | 2.6% | 15.4% | 2.18% | 2.94 |
| 12 | SixEleven | 7.7% | 7.7% | 15.4% | 0.0% | 7.7% | 1.34% | 2.5 |
| 13 | Select VoiceCom | 7.7% | 30.8% | 0.0% | 0.0% | 0.0% | 0.34% | 4.08 |
| 14 | Emapta | 7.1% | 0.0% | 7.7% | 7.7% | 12.8% | 0.17% | 2.09 |
| 15 | Booth & Partners | 5.8% | 7.7% | 0.0% | 7.7% | 7.7% | under 1% | 4.67 |
The finding that should worry marketers: the four assistants disagree
The industry talks about AI visibility as one number. It is four numbers, and for some firms they barely correlate.
| Company | ChatGPT | Claude | Gemini | Perplexity | Spread |
|---|---|---|---|---|---|
| Athena | 2.6% | 28.2% | 51.3% | 33.3% | 49 pts |
| ReceptionHQ | 2.4% | 19.0% | 14.3% | 42.9% | 40 pts |
| Moneypenny | 26.2% | 7.1% | 9.5% | 40.5% | 33 pts |
| AnswerFirst | 31.0% | 42.9% | 4.8% | 16.7% | 38 pts |
| SupportNinja | 38.5% | 15.4% | 2.6% | 7.7% | 36 pts |
| Select VoiceCom | 30.8% | 0.0% | 0.0% | 0.0% | 31 pts |
Athena is named in 51.3 percent of Gemini runs and 2.6 percent of ChatGPT runs. ReceptionHQ runs 42.9 percent on Perplexity against 2.4 percent on ChatGPT. Select VoiceCom exists almost exclusively inside ChatGPT. A firm measuring its AI visibility on one assistant is not measuring its AI visibility.
The AI receptionists are already inside the answering category
We tracked AI-native phone products separately rather than ranking them against human services. They no longer need the courtesy: 44.0 percent of all answering-service runs named at least one AI receptionist product. Goodcall appears in 45 runs, more than every human answering service below the top eight. When a buyer asks the “best answering service” question in 2026, the assistants routinely offer an AI product alongside the human ones.
| Product | Runs named | Share of all 480 runs |
|---|---|---|
| Goodcall | 45 | 9.4% |
| Upfirst | 21 | 4.4% |
| Allo | 20 | 4.2% |
| Slang.ai | 16 | 3.3% |
| My AI Front Desk | 4 | 0.8% |
| Aira | 2 | 0.4% |
Where the study author ranks
Armasourcing was named in 1 of 156 virtual assistant runs, which is 0.6 percent, by one assistant, once. In the 1,485 real user answers we found 2 mentions. We publish the index anyway, because the alternative is an industry measuring itself with numbers nobody can check. Our own climb, or failure to climb, will be visible in the Q4 edition like everyone else’s.
What actually drives these numbers
Every answer in the corpus carries its citations, so we can see what the assistants read before recommending anyone. The pattern is consistent: review platforms, Reddit threads and a handful of industry publishers supply the shortlists. The firms that dominate this index are the firms that dominate those sources, a dynamic we broke down in our guide to virtual assistant companies in the Philippines. For what the process of picking a provider should actually involve, start with how to choose a contact center outsourcing company.
Reuse and citation
All figures in this report may be reproduced with attribution and a link to this page. If you are citing it, this line works:
Armasourcing, The AI Visibility Index, Q3 2026, based on 1,485 real user AI answers and 480 controlled runs across ChatGPT, Claude, Gemini and Perplexity.
The full prompt list, so the panel can be replicated: best virtual assistant companies; best virtual assistant agency; top 10 virtual assistant companies; best virtual assistant services; affordable virtual assistant services; top real estate virtual assistant companies; best virtual assistant service for small business; best executive assistant service; which virtual assistant company should I hire for my business?; best virtual assistant company for customer service; best filipino virtual assistant companies; best virtual assistant companies in the philippines; where can I hire a full time virtual assistant from the philippines?; best answering service for small business; best answering services; affordable answering service; best virtual receptionist; what is the best phone answering service?; best medical answering service; best telephone answering services; affordable virtual receptionist; best virtual receptionist for small business; best legal answering services; best ai answering service; best 24/7 answering service; best after hours answering service; best virtual receptionist services; best call center company; what’s the best call center outsourcing company?; top bpo companies; top outsourcing companies; best outsourcing company; best customer service outsourcing companies; best customer support outsourcing for startups; affordable call center outsourcing for small business; best companies to outsource customer service to; best offshore staffing companies; call center outsourcing companies in the philippines; best bpo companies in the philippines for small business; best offshore staffing companies in the philippines.
Related first-party research: the Filipino Virtual Assistant Salary Report covers what candidates in this market ask to be paid, and the Customer Service Screening Report covers what unproctored hiring assessments actually measure. If you are structuring an offshore team, start at the contact center outsourcing hub or the Philippines contact center page, and measure it with the right contact center KPIs.
Frequently asked questions
Which company do AI assistants recommend most?
In this study, Smith.ai. It was named in 85.1 percent of 168 controlled answering-service runs across ChatGPT, Claude, Gemini and Perplexity, the highest rate for any company in any category. BELAY and Time Etc lead the virtual assistant category at 55.8 percent each, and Helpware leads SMB outsourcing at 39.7 percent.
Do ChatGPT, Claude, Gemini and Perplexity recommend the same companies?
Often not. Athena is named in 51.3 percent of Gemini runs but only 2.6 percent of ChatGPT runs. ReceptionHQ is strong on Perplexity and nearly absent from ChatGPT. Measuring AI visibility on a single assistant misses most of the picture.
How was the index measured?
Two sources, kept separate: 1,485 real user answers from ChatGPT and Google AI Overviews (DataForSEO LLM Mentions database, US, August 2025 to August 2026), and 480 controlled runs of 40 buyer questions across four assistants on August 3, 2026. A company counts only when named in the answer body with citation links stripped out, once per answer.
Why are Teleperformance, TaskUs and Accenture not ranked?
The index covers providers a small or mid-sized business can actually hire. Enterprise BPO and IT majors were excluded by rule, alongside freelance marketplaces like Upwork and Fiverr, which are platforms rather than managed providers. AI receptionist products are tracked in their own table rather than ranked against human services.
How often is the index updated?
Quarterly, with each edition’s dataset frozen on its collection date so editions can be compared. The Q4 2026 edition will report movement against the numbers on this page.
Need a VA who already understands your industry?
We don’t place generalists. Our VAs are matched and trained for the specific workflows of your sector.
See industry VAs →
